No full-text available
To read the full-text of this research,
you can request a copy directly from the authors.
Long-term learning in vernier acuity: Effects of stimulus orientation, range and of feedback
Abstract
In hyperacuity, as in many other tasks, performance improves with practice. To better understand the underlying mechanisms, we measured thresholds of 41 inexperienced observers for the discrimination of vernier displacements. In spite of considerable inter-individual differences, mean thresholds decreased monotonically over the 10,000 stimuli presented to each observer, if stimulus orientation was constant. Generalization of learning seemed to be possible across offset-ranges, but not across orientations. Learning was slightly faster with error feedback than without it in one experiment. These results effectively constrain the range of conceivable models for learning of hyperacuity.
... A fully stabilized or robust representation can be inferred when recognition performance has reached true asymptote. (Ramachandran & Braddick, 1973), spatial frequency waveform (Fiorentini & Berardi, 1980), vernier (Fahle & Edelman, 1993) and pop-out search (Ahissar & Hochstein, 1996) discrimination reveal remarkable improvements after extensive training with a specific stimulus set. However, this learning typically fails to generalize across very basic changes in stimulus orientation, size, or retinal position. ...
... In fact, overall processing rates for front, three-quarter, and profile views of self were generally comparable. These results contrast the highly specific learning effects found in visual discrimination studies that typically fail to generalize across simple changes in orientation, size, or retinal position (e.g., Ahissar & Hochstein, 1996;Fahle & Edelman, 1993;Fiorentini & Berardi, 1980;Ramachandran & Braddick, 1973). The fact that our observers could generalize to highly atypical views of their own face strongly supports the notion that robust representations contain some abstract or view-invariant information (Property 3). ...
... Indirect support of this notion comes from visual learning studies. When observers are extensively trained with a specific set of stimuli in psychophysical discrimination tasks, visual learning is dramatic but highly specific and typically fails to generalize to simple stimulus changes in size, orientation, or position (e.g., Ahissar & Hochstein, 1996;Fahle & Edelman, 1993;Fiorentini & Berardi, 1980;Ramachandran & Braddick, 1973). We therefore suspect that deriving some form of object constancy across stimulus change may be crucial for the development of a robust representation. ...
We report evidence from visual search that people can develop robust representations for highly overlearned faces. When observers searched for their own face versus the face of an unfamiliar observer, search slopes and intercepts revealed consistently faster processing of self than stranger. These processing advantages persisted even after hundreds of presentations of the unfamiliar face and even for atypical profile and upside-down views. Observers not only showed rapid asymptotic recognition of their own face as the target, but could reject their own face more quickly as the distractor. These findings suggest that robust representations for a highly overlearned face may (a) mediate rapid asymptotic visual processing, (b) require extensive experience to develop, (c) contain abstract or view-invariant information, (d) facilitate a variety of processes such as target recognition and distractor rejection, and (e) demand less attentional resources.
... VPL can be defined as a long-term performance enhancement on a visual task after visual experience (Seitz and Dinse, 2007;Sasaki et al., 2010;Sagi, 2011;Dosher and Lu, 2017). VPL is a powerful tool to improve visual detection and discrimination abilities in healthy subjects (e.g., Ball and Sekuler, 1982;Karni and Sagi, 1991;Fahle and Edelman, 1993;Bang et al., 2018) and in subjects with impaired vision (e.g., Polat et al., 2004;Ding and Levi, 2011;Hussain et al., 2012). VPL has been primarily investigated using primitive visual features such as texture (Karni and Sagi, 1991), orientation (Schoups et al., 2001) or motion direction (Ball and Sekuler, 1982;Frank et al., 2020b) but it can also occur for more complex visual stimuli such as visual feature conjunctions (e.g., Frank et al., 2014Frank et al., , 2016. ...
... Several factors modulate the development of VPL, including (but not limited to) (1) the total number of training sessions (e.g., Karni and Sagi, 1991), (2) the number of trials per training session (e.g., Amar-Halpert et al., 2017;Shmuel et al., 2020), (3) feedback signals during training (e.g., Fahle and Edelman, 1993;Herzog and Fahle, 1997;Frank et al., 2020a), (4) reinforcement signals during training (e.g., Seitz and Watanabe, 2003;Law and Gold, 2009;Roelfsema et al., 2010), (5) a night of continuous sleep (6-8 h) between successive training sessions (e.g., Tamaki et al., 2020), and (6) subject expertise prior to training (e.g., Bejjanki et al., 2014). We will briefly discuss the effectiveness of each factor in the Experimental design considerations. ...
... If explicit breaks are included in the training paradigm we recommend limiting the breaks to a few minutes to minimize the occurrence of stabilization of learning prior to the break, which could induce interference on new learning after the break. 3. Feedback signals during training a. Feedback about response accuracy speeds up VPL (Fahle and Edelman, 1993) and reduces variability in the development and speed of VPL between different subjects (Herzog and Fahle, 1997). For complex visual stimuli such as lesions in mammograms response feedback might even be necessary to develop long-lasting VPL (Frank et al., 2020a). ...
We describe a behavioral training protocol using visual perceptual learning (VPL) to improve visual detection skills in non-experts for subtle mammographic lesions indicative of breast cancer. This protocol can be adapted for the professional training of experts (radiologists) or to improve visual skills for other tasks, such as the detection of targets in photo or video surveillance.
For complete details on the use and execution of this protocol, please refer to Frank et al. (2020a).
... A main property of VPL is specificity, in which performance improvements are confined to the particular trained retinal location (Ball & Sekuler, 1982;Berardi & Fiorentini, 1987;Crist, Kapadia, Westheimer, & Gilbert, 1997;Dill & Fahle, 1997;Fahle & Edelman, 1993;Fahle, Edelman, & Poggio, 1995;Jehee, Ling, Swisher, van Bergen, & Tong, 2012;Schoups, Vogels, & Orban, 1995;Shiu & Pashler, 1992;Yang & Maunsell, 2004;Yashar et al., 2015), stimulus feature (Adab, Popivanov, Vanduffel, & Vogels, 2014;Adini, Sagi, & Tsodyks, 2002;Ahissar & Hochstein, 1997;Batson, Beer, Seitz, & Watanabe, 2011;Berardi & Fiorentini, 1987;Fiorentini & Berardi, 1980;Fiorentini & Berardi, 1981;Jehee et al., 2012;Watanabe, Náñez, & Sasaki, 2001;Yashar & Denison, 2017), or eye (Batson et al., 2011;Karni & Sagi, 1991). For example, monocular training on a particular line orientation in one quadrant of the visual field results in local accuracy improvements on that orientation for the trained eye, and no accuracy changes in the other three quadrants, or for the untrained eye, when the relative orientation of the line is orthogonal (Karni & Sagi, 1991, but see also Schoups & Orban, 1996, who found interocular transfer). ...
... Landolt-C acuity in the periphery has been shown to not improve with practice, possibly due to resolution constraints (Westheimer, 2001). Vernier offset discrimination thresholds improve with practice, but only for the trained orientation, location, or eye (Fahle, 2004;Fahle & Edelman, 1993;Fahle & Morgan, 1996;Fiorentini & Berardi, 1980;Poggio, Fahle, & Edelman, 1992; for reviews, see Sagi, 2011;Watanabe & Sasaki, 2015). ...
... In Experiment 2, for the neutral group, we found partial location transfer (some learning at the untrained location but not as much as at the trained location) and feature specificity (no learning for the untrained orientation). These results are consistent with previous studies of Vernier learning that, like our neutral condition, were performed under uncued or distributed attention, for which specificity regarding location, feature and eye was found (Fahle, 2004;Fahle & Edelman, 1993;Fahle et al., 1995). Notwithstanding such specificity, we found that training with valid, peripheral cues facilitates location transfer in the Vernier hyperacuity task, in which discrimination depends on fine spatial resolution. ...
Visual perceptual learning (VPL) refers to the improvement in performance on a visual task due to practice. A hallmark of VPL is specificity, as improvements are often confined to the trained retinal locations or stimulus features. We have previously found that exogenous (involuntary, stimulus-driven) and endogenous (voluntary, goal-driven) spatial attention can facilitate the transfer of VPL across locations in orientation discrimination tasks mediated by contrast sensitivity. Here, we investigated whether exogenous spatial attention can facilitate such transfer in acuity tasks that have been associated with higher specificity. We trained observers for 3 days (days 2-4) in a Landolt acuity task (Experiment 1) or a Vernier hyperacuity task (Experiment 2), with either exogenous precues (attention group) or neutral precues (neutral group). Importantly, during pre-tests (day 1) and post-tests (day 5), all observers were tested with neutral precues; thus, groups differed only in their attentional allocation during training. For the Landolt acuity task, we found evidence of location transfer in both the neutral and attention groups, suggesting weak location specificity of VPL. For the Vernier hyperacuity task, we found evidence of location and feature specificity in the neutral group, and learning transfer in the attention group-similar improvement at trained and untrained locations and features. Our results reveal that, when there is specificity in a perceptual acuity task, exogenous spatial attention can overcome that specificity and facilitate learning transfer to both untrained locations and features simultaneously with the same training. Thus, in addition to improving performance, exogenous attention generalizes perceptual learning across locations and features.
... For example, Fiorentini and Berardi (1980) demonstrated that perceptual learning is specific to both orientation and spatial frequency of the gratings used in a discrimination task. In a hyperacuity task, Poggio, Fahle, and Edelman (1992); Fahle and Edelman (1993) have shown that discriminative learning is highly specific to the location or orientation of stimulus in the visual field. Karni and Sagi (1991) have shown that in a visual texture discrimination task, learning is specific to the retina location of the stimulus, its orientation and also that there is no inter-ocular transfer of learning. ...
... Research on perceptual learning has recently focused on the extent to which perceptual learning is due to improvements in sensory abilities that are (informationally and temporally) earlier than the decision process itself or due to improvements in post-sensory and decision-related processing. Consistent with the former account, several psychophysics studies have demonstrated that perceptual learning is often highly specific to the location and other properties of the stimuli (Ball & Sekuler, 1987;Crist et al., 1997;Fahle & Edelman, 1993;Fiorentini & Berardi, 1980;Karni & Sagi, 1991;Poggio et al., 1992;Sagi & Tanne, 1994) implying specificity to the trained retinal location (Fahle, 2004(Fahle, , 2005. Similarly human FMRI studies offered evidence of activity enhancements in retinotopic areas corresponding to the trained visual fields (Schwartz et al., 2002) and increased responses along the whole hierarchy of early visual areas that correlated with improvements in behavioural performance following training over the course of several weeks (Furmanski et al., 2004;Jehee, Ling, Swisher, van Bergen, & Tong, 2012). ...
... The traditional perspective on the nature and locus of perceptual learning is that it arises early in the perceptual system. For example, well known seminal results from psychophysics have demonstrated that perceptual learning is often highly specific to the location and other properties of the stimuli (Fiorentini & Berardi, 1980;Poggio et al., 1992;Fahle & Edelman, 1993;Karni & Sagi, 1991;Crist et al., 1997;Ball & Sekuler, 1987;Sagi & Tanne, 1994), implying that its plasticity is highly specific to the trained retinal location (Fahle, 2004(Fahle, , 2005. In contrast to this traditional perspective, alternative psychophysical studies have proposed that perceptual learning arises not from plasticity in primary sensory areas, but rather from changes in how sensory signals are read out or interpreted by decision-making mechanisms Dosher and Lu (1999);Petrov et al. (2005); Lu et al. (2010). ...
Perceptual decision making involves the classification of sensory information, usually followed by an overt behavioural response. Any decision making, and perceptual decision making in particular, can be understood both theoretically and neurologically as a process of an accumulation of evidence to some threshold, at which point a commitment to a choice is made. This process can be examined in human subjects by analysing EEG data during perceptual decision making and identifying temporal components that are the neural signatures of the accumulation-to-bound decision making (see Philiastides & Sajda, 2006; Philiastides, Ratcliff, & Sajda, 2006a; Philiastides et al., 2006a; Ratcliff, Philiastides, & Sajda, 2009b). It may also be statistically modelled using sequential sampling models (see Ratcliff & Smith, 2004; Ratcliff, Gomez, & McKoon, 2004; Ratcliff & McKoon, 2008; Ratcliff & Van Dongen, 2011). Taken together, these provide us with a experimental and theoretical framework for the study of the neuroscience of human decision making. In this thesis, our aim is to address some open questions with respect to human perceptual decision making using the theoretical framework of sequential sampling models and the experimental paradigm of measuring temporal components in single-trial EEG discriminant analysis.
In Chapter 2, we will describe our studies of the role of learning on perceptual decision making. In particular, here we address competing hypotheses about the nature and location of perceptual learning in the brain. We provide evidence that perceptual learning arises from changes in higher level brain areas that are related to decision-making, rather than from perceptually earlier areas that are related to the encoding of sensory stimuli. In Chapter 3, we provide a specific mechanistic account of how learning affects perceptual decision making. This work follows from the work of others who have applied reinforcement learning theories (see Sutton & Barto, 1998) to the study of perceptual learning. In Chapter 4, we will describe our studies of the interaction of prior expectation and learning on decision making. In this study, we particularly aim to address whether prior expectation affects baseline activation or evidence accumulation in the decision making system, and how this changes with training. Here, we obtain evidence showing how the effects of prior expectation are more related to evidence accumulation rather than baseline activation. In Chapter 5, we provide sequential sampling models, particularly drift diffusion models, of the data that we’ve obtained in the main experiments described in Chapter 2 and Chapter 4. The principal results here show how learning and prior expectation primarily have their effect on perceptual decision making by increasing the rate of evidence accumulation.
Our general conclusion is that using a combination of the theoretical framework of sequential sampling models and the experimental paradigm of measuring temporal components in single-trial EEG discriminant analysis provides an effective and comprehensive means to address open questions with respect to human perceptual decision making.
... Interestingly, in studies that do report performance across several days of training (e.g., DeWind & Brannon, 2012;Park & Brannon, 2014 six training sessions), very little improvement in ANS ability was found after the first session (for a discussion of this pattern of results, see Lindskog & Winman, 2016). This lack of improvement after the first several hundred trials stands in stark contrast to most studies in the domain of perceptual learning, where improvements often continue for thousands or tens of thousands of trials (Fahle & Edelman, 1993;Larcombe, Kennard, & Bridge, 2017;Yang & Maunsell, 2004). ...
... In doing so, we made use of many of the commonly accepted best practices in the domain of perceptual learning. For instance, because one of the single best predictors of improvements on a perceptual learning task is simply time on task (as is true of most, if not all learning tasks - Badiru, 1992;Fahle & Edelman, 1993;Heathcote, Brown, & Mewhort, 2000), we made use of an extremely large number of training trials (8,000+). Indeed, given the shape of most perceptual learning curves, without employing a substantial number of training trials, it can often be quite difficult to determine whether learning has truly reached a hard asymptote at some final level of performance or instead is continuing to improve in the slow phase of an exponential function (Fahle & Edelman, 1993). ...
... For instance, because one of the single best predictors of improvements on a perceptual learning task is simply time on task (as is true of most, if not all learning tasks - Badiru, 1992;Fahle & Edelman, 1993;Heathcote, Brown, & Mewhort, 2000), we made use of an extremely large number of training trials (8,000+). Indeed, given the shape of most perceptual learning curves, without employing a substantial number of training trials, it can often be quite difficult to determine whether learning has truly reached a hard asymptote at some final level of performance or instead is continuing to improve in the slow phase of an exponential function (Fahle & Edelman, 1993). Furthermore, again consistent with best practices in training, practice was distributed through time (Larcombe et al., 2017), immediate informative feedback was provided (Fahle & Edelman, 1993;Herzog & Fahle, 1997;Shiu & Pashler, 1992), and task difficulty was modified throughout training to keep the task difficult, but doable (C. S. Green & Bavelier, 2008). ...
Recent research suggests that humans perceive quantity using a non-symbolic “number sense.” This sense is then thought to provide a foundation for understanding symbolic numbers in formal education. Given this link, there has been interest in the extent to which the approximate number system (ANS) can be improved via dedicated training, as this could provide a route to improving performance in symbolic mathematics. However, current evidence regarding the trainability of the ANS comes largely from studies that have used short training durations, leaving open the question of whether improvements occur over a longer time span. To address this limitation, we utilized a perceptual learning approach to investigate the extent to which long-term (8,000+ trials) training modifies the ANS. Consistent with the general methodological approach common in the domain of perceptual learning (where learning specificity is commonly observed), we also examined whether ANS training generalizes to: (a) untrained locations in the visual field; (b) an enumeration task; (c) a higher-level ratio comparison task; and (d) arithmetic ability. In contrast to previous short-term training studies showing that ANS learning quickly asymptotes, our long-term training approach revealed that performance continued to improve even after thousands of trials. We further found that the training generalized to untrained visual locations. At post-test there was non-significant evidence for generalization to a low-level enumeration task, but not to our high-level tasks, including ratio comparison, multi-object tracking, and arithmetic performance. These results demonstrate the potential utility of long-term psychophysical training, but also suggest that ANS training alone (even long-duration training) may be insufficient to modify higher-level math skills.
... We conceived an experimental procedure based on a perceptual task where the most prominent features of learning were operationalized as follows: (a) the procedure encompasses a temporal structure where a stimulus S D is given, the discrimination response R is provided by the user, which indicates the successful (or non-successful) outcome, and is followed by a feedback (see paragraph General structure of the training procedure); (b) the correct responses reflect the subject's capabilities of resolving the task over time [30], and the changes in task condition over time represent a learning curve (e.g. [19], [31], [32]); (c) we introduced a software framework, which allows the modulation of task difficulty (from easy to difficult, e.g., [32]) in a way that maximizes motivation and leads to a transition between different states in the learning process [33], [34]. ...
... Assuming that the training and the discrimination begins with easy task conditions, until the individual limit of the user's capability, and only subsequently proceeds to more difficult task conditions [33], we sought an estimate of this transition by the analysis of the history of the correct responses to the perceptual discrimination task. Afterwards, we verified the different states in the learning process by analyzing the changes of the task conditions over time, corresponding to the changes of the stimulus properties over time (e.g. the 'contrast threshold' in [32], or the 'Vernier acuity' in [31]), before and after the transition. We expected that positive changes over time in the learning curve before the transition (easy task conditions) would be a necessary condition to enable later learning effects on more difficult task conditions, reflected by positive changes in the learning curve over time after the transition. ...
Training programs, based on principles of brain-plasticity and skill learning, are useful in counteracting functional decline in pathological conditions. Training effects of such procedures are well described but their adaptive features are usually not reported. A software framework designed for a long-term home training program is presented. It gradually trains users, provides a multidimensional range of stimulus differentiation, encompasses a strategy to increase the task demand and includes motivational reinforcement components. The structured framework was tested in a feasibility study involving two perceptual discrimination tasks (visual and auditory) in four persons in middle-to-older adulthood who were trained for 30 days. Practicability of the training was shown in a home setting by high adherence to the procedure, adaptive increase in task demand over time and positive learning effects on an individual level. Participants learned to distinguish progressively smaller target objects in the visual task (with diminished contrast) and sweeps progressively varying less in frequency in the auditory task (with overlapping noise). This adaptive procedure can provide the basis for the design of extended training programs engaging sensory function in individuals with impaired sensorimotor and cognitive functions. Further investigations are necessary to assess the generalization of learning effects and clinical validity.
... In addition, recent research has shown that training can improve adults' sensitivity to fundamental dimensions of visual analysis (e.g., hyperacuity, orientation discrimination) on a long-term basis (for a recent review, see Kami & Bertini, 1997). Studies have shown that this enhanced sensitivity can start to develop quite rapidly (Fahle, Edelman, & Poggio, 1995) and that it may then continue to do so over weeks or months, during which participants complete many thousands of trials, even in the absence of feedback (e.g., Fahle & Edelman, 1993;Kami & Sagi, 1993). Of particular note is that learning is often highly specific to both the values on the stimulus dimensions and to the visual field position used during training. ...
... provements in sensitivity can start to occur very rapidly during a fast learning phase but also that they then continue to occur during a slow learning phase for extended periods of time (e.g., contrast Fahle et al., 1995, with Fahle & Edelman, 1993. The number of training trials we used was comparable to other work showing improvements in sensitivity during the fast phase of sensory learning (e.g., Fahle et al., 1995;Kami & Sagi, 1993). ...
Recent perceptual learning research has found long-term increases in the sensitivity of adults' perceptual systems. The authors examined whether such changes could partly explain improvement on tasks such as perception of medical X-ray images. Experiment 1 found experts' sensitivity to low-contrast dots in X-rays was better than novices'. Experiment 2 found a direction of luminance contrast-specific improvement in novices' detection of low-contrast dots in X-rays as a result of practice. Experiment 3 found a partly specific improvement in novices' detection of low-contrast features in real medical X-rays as a result of practice. Results suggest that experience enhances sensitivity to the critical dimensions of visual analysis for detecting abnormalities in X-ray images. Importantly, they demonstrate a real-world adaptive functional role for the long-term flexibility of sensory systems in adulthood.
... We conceived an experimental procedure based on a perceptual task where the most prominent features of learning were operationalized as follows: (a) the procedure encompasses a temporal structure where a stimulus S D is given, the discrimination response R is provided by the user, which indicates the successful (or non-successful) outcome, and is followed by a feedback (see paragraph General structure of the training procedure); (b) the correct responses reflect the subject's capabilities of resolving the task over time [28], and the changes in task condition over time represent a learning curve (e.g. [17], [29], [30]); (c) we introduced a software framework, which allows the modulation of task difficulty (from easy to difficult, e.g., [30]) in a way that maximizes motivation and leads to a transition between different states in the learning process [31], [32]. ...
... Assuming that the training and the discrimination begins with easy task conditions, until the individual limit of the user's capability, and only subsequently proceeds to more difficult task conditions [31], we sought an estimate of this transition by the analysis of the history of the correct responses to the perceptual discrimination task. Afterwards, we verified the different states in the learning process by analyzing the changes of the task conditions over time, corresponding to the changes of the stimulus properties over time (e.g. the 'contrast threshold' in [30], or the 'Vernier acuity' in [29]), before and after the transition. We expected that positive changes over time in the learning curve before the transition (easy task conditions) would be a necessary condition to enable later learning effects on more difficult task conditions, reflected by positive changes in the learning curve over time after the transition. ...
p>This preprint collection concerns the proof-of-principle study of an adaptive algorithm used in a long-term sensory home training. It consists of the description of the adaptive procedure and supplementary material to characterize the algorithm.
The main aim of the study was to determine whether the designed adaptive framework based on operant learning and brain-plasticity principles is able to promote learning over a long period of time, and whether indicators of learning can be derived from individual training data. The increase of task demands over time was shown in combination with positive learning effects at individual level. The study represents the proof-of-principle of the methods which can form the basis to design training procedures engaging simple and complex abilities such as sensory perception, sensory integration, motor control, and testing of complex psychometric functions.
</p
... We conceived an experimental procedure based on a perceptual task where the most prominent features of learning were operationalized as follows: (a) the procedure encompasses a temporal structure where a stimulus S D is given, the discrimination response R is provided by the user, which indicates the successful (or non-successful) outcome, and is followed by a feedback (see paragraph General structure of the training procedure); (b) the correct responses reflect the subject's capabilities of resolving the task over time [28], and the changes in task condition over time represent a learning curve (e.g. [17], [29], [30]); (c) we introduced a software framework, which allows the modulation of task difficulty (from easy to difficult, e.g., [30]) in a way that maximizes motivation and leads to a transition between different states in the learning process [31], [32]. ...
... Assuming that the training and the discrimination begins with easy task conditions, until the individual limit of the user's capability, and only subsequently proceeds to more difficult task conditions [31], we sought an estimate of this transition by the analysis of the history of the correct responses to the perceptual discrimination task. Afterwards, we verified the different states in the learning process by analyzing the changes of the task conditions over time, corresponding to the changes of the stimulus properties over time (e.g. the 'contrast threshold' in [30], or the 'Vernier acuity' in [29]), before and after the transition. We expected that positive changes over time in the learning curve before the transition (easy task conditions) would be a necessary condition to enable later learning effects on more difficult task conditions, reflected by positive changes in the learning curve over time after the transition. ...
p>This preprint collection concerns the proof-of-principle study of an adaptive algorithm used in a long-term sensory home training. It consists of the description of the adaptive procedure and supplementary material to characterize the algorithm.
The main aim of the study was to determine whether the designed adaptive framework based on operant learning and brain-plasticity principles is able to promote learning over a long period of time, and whether indicators of learning can be derived from individual training data. The increase of task demands over time was shown in combination with positive learning effects at individual level. The study represents the proof-of-principle of the methods which can form the basis to design training procedures engaging simple and complex abilities such as sensory perception, sensory integration, motor control, and testing of complex psychometric functions.
</p
... Le même résultat est trouvé en indiçage contextuel (Jiang & Chun, 2001). Cet apprentissage perceptif est très souvent hautement spécifique à un stimulus (Ahissar & Hochstein, 1993;Ball & Sekuler, 1987;Fahle, 1994;Fahle & Edelman, 1993;Fiorentini & Berardi, 1980) mais il a été montré que cela dépend de la difficulté de la tâche : ...
... D'autres études montrent qu'un feedback externe n'est pas nécessaire (e.g. Fahle & Edelman, 1993 (Croson & Sundali, 2005;Keren & Lewis, 1994;Sundali & Croson, 2006), aux loteries (Clotfelter & Cook, 1993), dans les paris sportifs (Alter & Oppenheimer, 2006) ou les courses hippiques (Terrell, 1994). Il connait deux formes : ...
La bistabilité est une surprenante alternance de l’apparence d’un stimulus entre deux interprétations de ce stimulus. Elle a lieu lorsqu’une stimulation physique est très ambigüe, comme lorsque l’on présente une image dans un oeil très différente de l’image de l’autre oeil (rivalité binoculaire). La bistabilité implique de (1) décider que le stimulus est bistable ; (2) sélectionner le premier percept ; (3) supprimer l’autre percept ; (4) décider quand alterner. Tandis que les deux dernières étapes sont très étudiées, les deux premières étapes restent encore mal connues. Ici, nous approfondissons les connaissances sur les mécanismes qui décident la rivalité plutôt que la fusion (étude 4), sur le choix du premier percept (études 1 à 5) et sur la dynamique bistable (étude 1 et 3). Notamment, la littérature actuelle favorise une explication bas-niveau de la rivalité binoculaire, c’est-à-dire ayant lieu dès les premières étapes de traitement cognitif, avant tout traitement complexe. Le but principal de cette thèse est de clarifier l’influence potentielle sur la rivalité de traitements probabilistes et donc plus complexes.J’ai étudié (1) l’influence de l’utilité d’une interprétation sur la bistabilité ; (2) la manière de changer les préférences bistables ; (3) la spécificité du ou des premiers percepts ; (4) l’influence d’illusions d’orientations sur la stéréo-fusion et la rivalité ; (5) l’influence du passé sur le percept bistable, sous la forme de prédictions du système.J’ai trouvé :(1) qu’une interprétation utile est perçue plus souvent sur le premier percept d’un épisode de rivalité binoculaire. Cette influence est implicite, non-attentionnelle et implique des calculs probabilistes sur l’utilité d’un percept.(2) que l’utilité peut modifier les préférences en rivalité de transparence de mouvement et que ces préférences ont donc pour rôle de favoriser la réussite à une tâche.(3) que le premier percept de la rivalité binoculaire n’est pas le seul à être différent des autres et que le contrôle volontaire est plus fort au début d’un épisode bistable pour le percept non-favorisé.(4) que l’étape de contraste illusoire entre orientations occupe une position hiérarchique plus basse que les étapes de vision stéréoscopique et de rivalité.(5) une forte corrélation positive entre le passé visuel ancien et l’orientation perçue en rivalité binoculaire. Nous expliquons ce résultat par un modèle d’adaptation prédictive selon lequel le système visuel prédit le prochain percept tel que la distribution récente des orientations corresponde à une distribution plus ancienne. Ici encore, des calculs probabilistes sont impliqués.En résumé, ce travail démontre principalement l’existence de traitements probabilistes en perception ambigüe, par l’intermédiaire d’un calcul d’utilité et de prédictions dans l’adaptation.
... Liu, Lu, & Dosher, 2010Z. Liu & Weinshall, 2000), training schedule (Hung & Seitz, 2014;Xiao et al., 2008), reward (Seitz, Kim, & Watanabe, 2009;Zhang et al., 2018), feedback (Aberg & Herzog, 2012;Fahle & Edelman, 1993;J. Liu et al., 2010;Shibata, Yamagishi, Ishii, & Kawato, 2009), and attention (Donovan, Szpiro, & Carrasco, 2015;Mukai et al., 2007) conditions were also evaluated. ...
... To construct a learning curve, it is necessary to select a proper algorithm to measure thresholds during training. In the field of perceptual learning, the up-down staircase method is often used to estimate the threshold for a certain percent of correct responses by adjusting the stimulus level (e.g., contrast, luminance, orientation difference, and motion coherence) based on the subject's responses (Fahle & Edelman, 1993;Xie & Yu, 2018;Yehezkel, Sterkin, Lev, Levi, & Polat, 2016;Yu, Klein, & Levi, 2004;Zhang et al., 2018). Before training, the experimenter often determined an initial stimulus level based on the pilot data. ...
Computerized cognitive training (CCT) has been found to improve a range of skills such as attention, working memory, inhibition control, and decision making. However, the relationship between the initial performance, amount of improvement, time constant, and asymptotic performance level in CCT is still unclear. In the current study, we performed selective attention training on college students and addressed this issue by mathematically modeling the learning curve with an exponential function. Twenty-nine students completed approximately 10 days of CCT. Presentation time served as the dependent variable and was measured by three-down/one-up adaptive algorithms. We fitted an exponential function to the estimated block thresholds during CCT and obtained three learning parameters (amount of improvement, time constant, and asymptotic performance level) for all subjects. The initial performance was defined by the sum of the amount of improvement and the asymptotic performance level. Pearson correlation analyses were conducted between the initial performance and the three leaning parameters. The initial performance was positively correlated with the amount of improvement and asymptotic performance level, but was negatively correlated with the time constant. The time constant was negatively correlated with the amount of improvement and asymptotic performance level. Poorer initial performance was linked to a larger amount of improvement, shorter time constant, and higher asymptotic threshold, which supported the compensation account. Our results may help improve the present understanding of the nature of the CCT process and demonstrate the advantages of using a customized training protocol to enhance the efficiency of cognitive training in practical applications.
... Learning has also been found in the absence of external feedback (Ball & Sekuler, 1982;Fahle, 2005;Liu et al., 2012;Petrov et al., 2006;Vaina et al., 1998;Shiu & Pashler, 1992), specifically when the task is easy, but not when only difficult trials are presented (Liu et al., 2010). While learning without feedback has been found when easy and difficult trials are interleaved (Fahle & Edelman, 1993;Liu et al., 2012), this is not always the case (Seitz et al., 2006;Asher et al., 2019). ...
... When external, trial-by-trial feedback is provided, this may be used to guide the learning process. In the absence of external feedback, it has been proposed that easy trials can act to 'bootstrap' the more difficult trials (Fahle & Edelman, 1993). Petrov et al. (2005) proposed that, in the absence of external feedback, weights are updated under the assumption that the response is accurate. ...
... For instance, in the visual domain, training can improve performance in Vernier acuity tasks (Duckman, 2006, pp. 36-38), but this hyperacuity does not transfer to three-dot bisection tasks (Fahle & Morgan, 1996), monocular contralateralization, or simple rotation of the stimulus (Fahle & Edelman, 1993). Similarly, in olfactory discrimination tasks, changing the odors serving as the "background" from which the target is selected can undo the practice-based boost in identification accuracy (Wilson & Stevenson, 2003). ...
... Although this finding is consistent with some studies that tout training generalizability (e.g., Ragert et al., 2004), many others report highly-specific sensitivity improvements (e.g., Duckman, 2006, pp. 36-38;Fahle & Edelman, 1993;Fahle & Morgan, 1996). This may be due in part to the precision requirements of the task. ...
The present study explored whether timbre discrimination could be honed through perceptual exposure training. A timbre continuum that blended the timbres of an oboe and a trumpet was constructed to produce novel timbres to which participants had no previous exposure. Participants' baseline sensitivity (d’) in discriminating tones comprising these novel timbres were compared to their sensitivity after training. In Experiments 1 and 2, training involved exposure to a tone discrimination task, with some participants getting feedback about the accuracy of their responses and other participants getting no feedback. Although there was no improvement with initially sub-threshold timbral variation regardless of whether there was feedback or not after each exposure, exposure with feedback did improve discrimination with initially above-threshold timbral variation. Experiment 3 showed that this improved timbre discrimination via exposure with feedback also helped segregate a melody from a set of jumbled tones, and, moreover, generalized to the segregation of a new melody played in the trained timbre. The results are discussed within the framework of perceptual learning.
... The VAmeasures we obtained correspond well to previous reports investigating part of the different conditions only in isolation. For photopic hVA, we report an average of − 1.16 LogMAR (14.45 decimal acuity), which corresponds to previous studies ranging between − 1.47 and − 0.87 LogMAR [6,[33][34][35]. For scotopic cVA, we report an average of 0.91 LogMAR (0.12 decimal acuity), which corresponds to previous reports of the same cVA value [0.92 LogMAR (0.12 decimal acuity)] for a luminance of 3 × 10 − 3 cd/m 2 [11][12][13]36] and a range of 0.79 to 1.76 LogMAR for a lower luminance (9 × 10 − 4 cd/m 2 ; For scotopic hVA, we report an average of − 0.17 LogMAR (1.47 decimal acuity) i.e. better 1 acuities than a previous study investigating two participants with a hVA of 0.22 LogMAR [4]. ...
... We did not observe a significant intersession effect for any of the acuity types tested. Hence, the well-known training effects reported for photopic hVA [35,[37][38][39] and also cVA [40], need more repetitions than single measurement runs taken in two different session [30]. In fact, the previously reported training effects were observed after hundreds of trials. ...
Purpose
Visual acuity (VA) is an important determinant of visual function. Here we establish procedures and recommendations for VA testing extending beyond the classical VA and thus make them available for future studies of visual function in health and disease. Specifically, we provide reference values for photopic and scotopic conventional uncrowded visual acuity (cVA) and Vernier-hyperacuity (hVA) and assess their reproducibility and dependence on contrast polarity.
Methods
For ten observers with normal vision, we determined photopic (“p”; maximal luminance 220 cd/m²) and scotopic (“s”; maximal luminance 0.004 cd/m²; 40 min of dark adaptation) cVA and hVA, for two contrast polarities i.e. black optotypes on white background and vice versa. To assess intersession effects, two sets of measurements were obtained on different days.
Results
Compared to pcVA (1.32 decimal VA; − 0.12 ± 0.02 LogMAR), the phVA (14.45 decimal VA; − 1.16 ± 0.04 LogMAR) scaled (in terms of decimal visual acuity) on average with a factor 11.0, the scVA (0.12 decimal VA; 0.91 ± 0.03 LogMAR) with a factor of 0.1, and the shVA (1.47 decimal VA; − 0.17 ± 0.02 LogMAR) with a factor of 1.1. There were neither significant effects of contrast polarity (p > 0.12), nor of session (p > 0.28).
Conclusions
Our approach optimises integrated photopic and scotopic cVA and hVA measurements for general use and thus encourages the integration of these important measures of scotopic visual function in future studies. The absence of strong intersession effects demonstrates that no dedicated training session is needed to obtain scotopic and hVA measurements. The combined measures of scotopic and photopic VAs open a field of applications to study interplay and plasticity of the retinal photoreceptor systems and cortical processing in health and visual disease. As a rule of thumb, hyperacuity is 10× higher both in the photopic and scotopic range than conventional acuity. Thus, scotopic hyperacuity is close to photopic conventional acuity.
... Perceptual learning refers to any relatively permanent and consistent change in the perception of objects and their features, following the training of the object and its features (Gibson, 1963). A large number of previous studies have shown that perceptual learning can improve the detection and discrimination of many basic visual features, such as orientation (Hu et al., 2021;Zhang et al., 2023), contrast (Yu et al., 2004;Zhou et al., 2006;Hua et al., 2010;Yu et al., 2016;Roberts and Carrasco, 2022), spatial phase (Berardi and Fiorentini, 1987), stereoacuity (Fendick and Westheimer, 1983;Roberts and Carrasco, 2022), visual acuity (Zhou et al., 2006;Eisen-Enosh et al., 2023), vernier acuity (Fahle and Edelman, 1993), and texture (Karni and Sagi, 1991). Perceptual learning has long been characterized by its specificity to learned features. ...
Introduction
Perceptual learning of facial expression is shown specific to the train expression, indicating separate encoding of the emotional contents in different expressions. However, little is known about the specificity of emotional recognition training with the visual search paradigm and the sensitivity of learning to near-threshold stimuli.
Methods
In the present study, we adopted a visual search paradigm to measure the recognition of facial expressions. In Experiment 1 (Exp1), Experiment 2 (Exp2), and Experiment 3 (Exp3), subjects were trained for 8 days to search for a target expression in an array of faces presented for 950 ms, 350 ms, and 50 ms, respectively. In Experiment 4 (Exp4), we trained subjects to search for a target of a triangle, and tested them with the task of facial expression search. Before and after the training, subjects were tested on the trained and untrained facial expressions which were presented for 950 ms, 650 ms, 350 ms, or 50 ms.
Results
The results showed that training led to large improvements in the recognition of facial emotions only if the faces were presented long enough (Exp1: 85.89%; Exp2: 46.05%). Furthermore, the training effect could transfer to the untrained expression. However, when the faces were presented briefly (Exp3), the training effect was small (6.38%). In Exp4, the results indicated that the training effect could not transfer across categories.
Discussion
Our findings revealed cross-emotion transfer for facial expression recognition training in a visual search task. In addition, learning hardly affects the recognition of near-threshold expressions.
... Previous studies on various learning types (e.g., habituation, perceptual learning) have commonly applied this simple experimental procedure [1][2][3]. In the vision domain, numerous investigations demonstrated that learning can occur in response to mere exposure to repetitive stimulation without any explicit training (i.e., exposurebased learning), even only with mental imagery in the absence of stimulus exposure for various visual performances, including motion direction discrimination [4][5][6][7]. It has been shown that such sensory learning induces rapid recalibration of visual processing and leads to lasting changes in perception and goal-directed behavior [8,9]. ...
... Visual perceptual learning can occur at both lowand high-levels of the visual processing hierarchy in the brain (Op de Beeck & Baker, 2010;Watanabe & Sasaki, 2015). For example, many studies have shown that training remarkably enhances visual abilities to detect or discriminate low-level visual features such as contrast (Dorais & Sagi, 1997;Yu, Zhang, Qiu, & Fang, 2016), orientation (Schoups, Vogels, & Orban, 1995), spatial frequency (Fiorentini & Berardi, 1980), spatial phase (Berardi & Fiorentini, 1987), and hyperacuity (Fahle & Edelman, 1993). Similarly, recognition and discrimination of high-level visual stimuli, such as shape (Kourtzi, Betts, Sarkheil, & Welchman, 2005), object (Furmanski & Engel, 2000;Sigman & Gilbert, 2000), and face (Bi, Chen, Weng, He, & Fang, 2010), can be substantially improved by training, as well. ...
Visual perceptual learning has been studied extensively and reported to enhance the perception of almost all types of training stimuli, from low- to high-level visual stimuli. Notably, high-level stimuli are often composed of multiple low-level features. Therefore, it is natural to ask whether training of high-level stimuli affects the perception of low-level stimuli and vice versa. In the present study, we trained subjects with either a high-level configuration stimulus or a low-level element stimulus. The high-level configuration stimulus consisted of two Gabors in the left and right visual fields, respectively, and the low-level element stimulus was the Gabor in the right visual field of the configuration stimulus. We measured the perceptual learning effects using the configuration stimulus and the element stimuli in both left and right visual fields. We found that the configuration perceptual learning equally improved the perception of the configuration stimulus and both element stimuli. In contrast, the element perceptual learning was confined to the trained element stimulus. These findings demonstrate an asymmetric relationship between perceptual learning of the configuration and the element stimuli and suggest a hybrid mechanism of the configuration perceptual learning. Our findings also offer a promising paradigm to promote the efficiency of perceptual learning-that is, gaining more learning effect with less training time.
... Feedback and reward also influence perceptual learning. Learning can occur with block feedback, which only informs the observer of their aggregate performance accuracy over many trials 44,45 , or without any feedback [44][45][46][47][48][49] . However, trial-by-trial feedback can improve the rate of learning or enable learning in otherwise not learnable tasks 50,51 . ...
The visual expertise of adult humans is jointly determined by evolution, visual development and visual perceptual learning. Perceptual learning refers to performance improvements in perceptual tasks after practice or training in the task. It occurs in almost all visual tasks, ranging from simple feature detection to complex scene analysis. In this Review, we focus on key behavioural aspects of visual perceptual learning. We begin by describing visual perceptual learning tasks and manipulations that influence the magnitude of learning, and then discuss the specificity of learning. Next, we present theories and computational models of learning and specificity. We then review applications of visual perceptual learning in visual rehabilitation. Finally, we summarize the general principles of visual perceptual learning, discuss the tension between plasticity and stability, and conclude with new research directions. Perceptual learning, or performance improvements after training on perceptual tasks, is a widespread phenomenon in visual perception. In this Review, Lu and Dosher describe findings regarding the specificity and transfer of perceptual learning, mechanisms of learning and key applications in visual rehabilitation.
... In fact, reports in the literature on adult perceptual learning indicate that improvements in performance following training are task or stimulus-specific. For instance, the learning effect in simple detection and discrimination tasks is specific to the learned orientation of the stimulus (Fahle and Edelman, 1993;Fiorentini and Berardi, 1980;Poggio, Fahle and Edelman, 1992), the spatial frequency of the stimulus Berardi, 1980,1981), and the direction of stimulus motion Sekuler, 1982, 1987). Furthermore, the degree of transfer of the learning effect to other tasks, to the same task performed in a different retinal location, or with a different eye is still debated, with estimates of transfer ranging from 0% to 100% (Beard, Levi and Reich, 1995;Berardi, 1980, 1981;Ball and Sekuler, 1987). ...
Visual and linguistic factors in literacy acquisition: Instructional Implications For Beginning Readers in Low-Income Countries. A literature review prepared for the Global Partnership for Education, c/o World Bank.
... In perceptual learning, it is often the case that improvements are specific to the trained stimulus and do not tend to generalize. Early studies of transfer of perceptual learning found it to be highly specific for retinal position (Fahle et al., 1995;Karni & Sagi, 1991), orientation (Fahle & Edelman, 1993;Vogels & Orban, 1985), and trained eye (Karni & Sagi, 1991), with no transfer between similar tasks. ...
The material in The Cognitive Unconscious began as a master’s thesis that examined the manner in which knowledge of fairly complex, patterned material could be acquired without any conscious effort to learn it and with little to no awareness of what had been learned. It was dubbed implicit learning and, over more than fifty years, became a vigorously researched area in the social sciences. The Cognitive Unconscious brings together several dozen scientists from a variety of backgrounds and presents a broad (and deep) overview of how the exploration of the cognitive unconscious grew from that first study to a domain of research to which contributions have been made by sociologists, neuroscientists, evolutionary biologists, modelers, social and organizational psychologists, sport psychologists, primatologists, developmentalists, linguists, psychiatrists and psychotherapists, and measurement and assessment researchers. The core message seems fairly straightforward. Unconscious, implicit cognitive processes play a role in virtually everything interesting that human beings do. The implicit and explicit elements of cognition form a rich and complex interactive framework that make up who we are. The volume has contributions from over thirty distinguished authors from nine different countries and gives a balanced and thorough overview of where the field is today, a bit over a half-century since the first experiments were run.
... In perceptual learning, it is often the case that improvements are specific to the trained stimulus and do not tend to generalize. Early studies of transfer of perceptual learning found it to be highly specific for retinal position (Fahle et al., 1995;Karni & Sagi, 1991), orientation (Fahle & Edelman, 1993;Vogels & Orban, 1985), and trained eye (Karni & Sagi, 1991), with no transfer between similar tasks. ...
The material in The Cognitive Unconscious began as a master’s thesis that examined the manner in which knowledge of fairly complex, patterned material could be acquired without any conscious effort to learn it and with little to no awareness of what had been learned. It was dubbed implicit learning and, over more than fifty years, became a vigorously researched area in the social sciences. The Cognitive Unconscious brings together several dozen scientists from a variety of backgrounds and presents a broad (and deep) overview of how the exploration of the cognitive unconscious grew from that first study to a domain of research to which contributions have been made by sociologists, neuroscientists, evolutionary biologists, modelers, social and organizational psychologists, sport psychologists, primatologists, developmentalists, linguists, psychiatrists and psychotherapists, and measurement and assessment researchers. The core message seems fairly straightforward. Unconscious, implicit cognitive processes play a role in virtually everything interesting that human beings do. The implicit and explicit elements of cognition form a rich and complex interactive framework that make up who we are. The volume has contributions from over thirty distinguished authors from nine different countries and gives a balanced and thorough overview of where the field is today, a bit over a half-century since the first experiments were run.
... Relatively poor agreement between individual measurements of stereoacuity across different clinical stereotests has been established. 43 Moreover, like other visual tasks such as motion discrimination, 44 luminance contrast detection, 45 texture discrimination 46 and other hyperacuity tasks such as Vernier acuity [47][48][49] and line orientation discrimination, 50 stereopsis can be improved through repeated practice over the course of hundreds and sometimes thousands of trials in psychophysical experiments. [51][52][53][54][55] Perceptual learning in stereoacuity occurs both with 56 and without 57 feedback on performance, although stronger learning effects occur when feedback is given. ...
Introduction:
Stereoacuity, like many forms of hyperacuity, improves with practice. We investigated the effects of repeated measurements over multiple visits on stereoacuity using two commonly utilised clinical stereotests, for both crossed and uncrossed disparity stimuli.
Methods:
Participants were adults with normal binocular vision (n = 17) aged between 18 and 50 years. Stereoacuity was measured using the Randot and TNO stereotests on five separate occasions over a six week period. We utilised both crossed and uncrossed stimuli to separately evaluate stereoacuity in both disparity directions. A subset of the subject group also completed a further five visits over an additional six week period. Threshold stereoacuity was determined by the lowest disparity level at which the subjects could correctly identify both the position and disparity direction (crossed or uncrossed) of the stimulus. Data were analysed by repeated measures analysis of variance.
Results:
Stereoacuity for crossed and uncrossed stimuli improved significantly across the first five visits (F1,21 = 4.24, p = 0.05). The main effect of disparity direction on stereoacuity was not significant (F1 = 0.02, p = 0.91). However, a significant interaction between disparity direction and stereotest was identified (F1 = 7.92, p = 0.01).
Conclusions:
Stereoacuity measured with both the TNO and Randot stereotests improved significantly over the course of five repetitions. Although differences between crossed and uncrossed stereoacuity were evident, they depended on the stereotest used and reduced or disappeared after repeated measurements. A single measure of stereoacuity is inadequate for properly evaluating adult stereopsis clinically.
... Indeed, mere exposure to visual stimuli can lead to perceptual learning [14]. It has been known since the landmark work of Gibson and Gibson in the 1950s that perceptual learning can occur without feedback [15]; practice and exposure without supervision (in the machine learning sense) can also result in perceptual learning [10,16]. ...
When searching a visual image that contains multiple target objects of interest, human subjects often show a satisfaction of search (SOS) effect, whereby if the subjects find one target, they are less likely to find additional targets in the image. Reducing SOS or, equivalently, subsequent search miss (SSM), is of great significance in many real-world situations where it is of paramount importance to find all targets in a given image, not just one. However, studies have shown that even highly trained and experienced subjects, such as expert radiologists, are subject to SOS. Here, using the detection of camouflaged objects (or camouflage-breaking) as an illustrative case, we demonstrate that when naïve subjects are trained to detect camouflaged objects more effectively, it has the side effect of reducing subjects’ SOS. We tested subjects in the SOS task before and after they were trained in camouflage-breaking. During SOS testing, subjects viewed naturalistic scenes that contained zero, one, or two targets, depending on the image. As expected, before camouflage-training, subjects showed a strong SOS effect, whereby if they had found a target with relatively high visual saliency in a given image, they were less likely to have also found a lower-saliency target when one existed in the image. Subjects were then trained in the camouflage-breaking task to criterion using non-SOS images, i.e., camouflage images that contained zero or one target. Surprisingly, the trained subjects no longer showed significant levels of SOS. This reduction was specific to the particular background texture in which the subjects received camouflage training; subjects continued to show significant SOS when tested using a different background texture in which they did not receive camouflage training. A separate experiment showed that the reduction in SOS was not attributable to non-specific exposure or practice effects. Together, our results demonstrate that perceptual expertise can, in principle, reduce SOS, even when the perceptual training does not specifically target SOS reduction.
... It is the ability to detect the slight horizontal misalignment of two vertical lines (or bars). Vernier acuity can be as low as 3 second of arc (arc sec) in individuals who have had extensive practice and approximately 20 arc sec for naïve subjects [77,78]. ...
Freezing of gait (FOG) is considered to be a motor disorder symptom that affects some Parkinson Disease (PD) patients; however, it is hypothesized that sensory systems may also be involved in FOG. This review article summarizes the results from previous studies focusing on visual functions in PD patients. More emphasize will be focused on freezing of gait PD patients and whether visual functions are affected to greater amount among them than non-freezing of gait PD patients. Visual functions include high contrast visual acuity, low contrast visual acuity, contrast sensitivity, Vernier acuity, mesopic vision, stereopsis, motion perception, and vergence eye movements are all affected in PD patients, with FOG patients having more deficits in some of these functions. FOG patients also had larger impairments in non-dopaminergic mediated functions such as pupil light reflex and visual processing speed test, which suggests that FOG patients have greater impairment in two functions that involve cholinergic neurotransmitters. Whether these impairments are contributing to the FOG or just associated with FOG is uncertain.
... In addition, it is conceivable that similar to ''fast learning'' mechanisms that are engaged in initial training, 6,37 memory reactivations may enable ASD participants to learn general aspects of the task, possibly by engaging higher-order regions that communicate with early visual areas that encode the task. 38 In turn, this may facilitate generalization of learning, which is absent in extensive training conditions, 15,39 showing high specificity to trained stimulus features 16,[40][41][42][43][44][45] (but see single condition in Harris et al. 4 ). The relation between these mechanisms and reactivationinduced learning in ASD remains to be determined. ...
Visual skill learning is the process of improving responses to surrounding visual stimuli.¹ For individuals with autism spectrum disorders (ASDs), efficient skill learning may be especially valuable due to potential difficulties with sensory processing² and challenges in adjusting flexibly to changing environments.³,⁴ Standard skill learning protocols require extensive practice with multiple stimulus repetitions,5, 6, 7 which may be difficult for individuals with ASD and create abnormally specific learning with poor ability to generalize.⁴ Motivated by findings indicating that brief memory reactivations can facilitate skill learning,⁸,⁹ we hypothesized that reactivation learning with few stimulus repetitions will enable efficient learning in individuals with ASD, similar to their learning with standard extensive practice protocols used in previous studies.⁴,¹⁰,¹¹ We further hypothesized that in contrast to experience-dependent plasticity often resulting in specificity, reactivation-induced learning would enable generalization patterns in ASD. To test our hypotheses, high-functioning adults with ASD underwent brief reactivations of an encoded visual learning task, consisting of only 5 trials each instead of hundreds. Remarkably, individuals with ASD improved their visual discrimination ability in the task substantially, demonstrating successful learning. Furthermore, individuals with ASD generalized learning to an untrained visual location, indicating a unique benefit of reactivation learning mechanisms for ASD individuals. Finally, an additional experiment showed that without memory reactivations ASD subjects did not demonstrate efficient learning and generalization patterns. Taken together, the results provide proof-of-concept evidence supporting a distinct route for efficient visual learning and generalization in ASD, which may be beneficial for skill learning in other sensory and motor domains.
... were rotated relative to each other across the four quadrants and each stimulus was presented with contrast jittered between 70%, 80%, 90%, and 100% of its original value. This was done to prevent learning effects or the use of image matching strategies based on factors other than temporal slope; these techniques have been found to influence sensitivity scores (Fahle & Edelman, 1993;Herzog & Fahle, 1997). ...
Natural scenes contain several statistical regularities despite their superficially diverse appearances (e.g., mountains, rainforests, deserts). First, they exhibit a unique distribution of luminance intensities decreasing across spatial frequency, known as the 1/fα amplitude spectrum (α ≈ 1). Additionally, natural scenes share consistent geometric properties, comprising similar densities of structure across multiple scales-a property classifying them as fractal (e.g., how the branching patterns of rivers and trees appear similar irrespective of scale). These two properties are intimately related and correlate strongly in natural scenes. However, research using thresholded noise images suggests that spatially, the human visual system is preferentially tuned to natural scene structure more so than 1/fα spectra. It is currently unclear whether this dependency on natural geometry extends to the temporal domain. We used a psychophysics task to measure discrimination sensitivity toward two types of synthetic noise movies: gray scale and thresholded (N = 60). Each movie type shared the same geometric properties (measured fractal D), but substantially differing spectral properties (measured α). In both space and time, we observe a characteristic dependency on stimulus structure across movie types, with sensitivity peaking for stimuli with natural geometry despite having altered 1/fα spectra. Although only measured behaviorally, our findings may imply that the neural processes underlying this tuning have developed to be sensitive to the most stable signal in our natural environment-structure (e.g., the structural properties of a tree are consistent from morning to night despite illumination changes across time points).
... These processes contribute to a range of complex behaviors that are important for typical development, such as face processing [6], reading [7], occupational expertise [8,9], and higher cognition [5,10]. Changes associated with learning are often the result of extensive experience [11][12][13][14] but may also occur quite quickly [2,15] However, little is understood about how internal states (e.g., emotional valence or arousal) influence these learning processes, despite the established neural links between certain affective and learning processes (e.g., links between reward and learning networks, [16]; common stress neuromodulators of hippocampal and amygdalar function [17,18]). Here, we combine perspectives from perceptual, affective, and developmental sciences to assess whether components of learning are influenced by manipulations of either valence (i.e., positive vs. negative mood manipulation) or arousal (manipulated vs. non-manipulated mood). ...
Despite clear links between affective processes in many areas of cognition and perception, the influence of affective valence and arousal on low-level perceptual learning have remained largely unexplored. Such influences could have the potential to disrupt or enhance learning that would have long-term consequences for young learners. The current study manipulated 8- to 11-year-old children's and young adults' mood using video clips (to induce a positive mood) or a psychosocial stressor (to induce a negative mood). Each participant then completed one session of a low-level visual learning task (visual texture paradigm). Using novel computational methods, we did not observe evidence for the modulation of visual perceptual learning by manipulations of emotional arousal or valence in either children or adults. The majority of results supported a model of perceptual learning that is overwhelmingly constrained to the task itself and independent from external factors such as variations in learners' affect.
... While PL transfers to stimuli highly similar to θ tr as expected, it causes performance for intermediate stimuli ("proximal", Fig.7) to drop below pre-PL levels. Indeed, some experiments report worse-than-baseline performance when subjects are tested on untrained stimuli following PL [55,56,57]. In addition, PL transfers to stimuli further away ("distal") from θ tr . ...
Perceptual learning (PL) involves long-lasting improvement in perceptual tasks following extensive training. Such improvement has been found to correlate with modifications in neuronal response properties in early as well as late sensory cortical areas. A major challenge is to dissect the causal relation between modification of the neural circuits and the behavioral changes. Previous theoretical and computational studies of PL have largely focused on single-layer model networks, and thus did not address salient characteristics of PL arising from the multiple-staged "deep" structure of the perceptual system. Here we develop a theory of PL in a deep neuronal network architecture, addressing the questions of how changes induced by PL are distributed across the multiple stages of cortex, and how do the respective changes determine the performance in fine discrimination tasks. We prove that in such tasks, modifications of synaptic weights of early sensory areas are both sufficient and necessary for PL. In addition, optimal synaptic weights in the deep network are not unique but span a large space of solutions. We postulate that, in the brain, plasticity throughout the deep network is distributed such that the resultant perturbation on prior circuit structures is minimized. In contrast to most previous models of PL, the minimum perturbation (MP) learning does not change the network readout weights. Our results provide mechanistic and normative explanations for several important physiological features of PL and reconcile apparently contradictory psychophysical findings.
... However, although feedback is considered essential for various types of learning, perceptual learning studies have reported that participants can improve their performance in visual perceptual tasks without performance feedback, such as motion-direction discrimination task 37 and texture discrimination task 38 . Researchers even have found that the learning rate is similar to and without feedback in a direction discrimination task 39 . These perceptual improvements are generally attributed to the neural plasticity at the cellular level in the visual system 40 . ...
Accumulating evidence indicates that the spatial error of human's hand localization appears subject-specific. However, whether the idiosyncratic pattern persists across time with good within-subject consistency has not been adequately examined. Here we measured the hand localization map by a Visual-matching task in multiple sessions over 2 days. Interestingly, we found that participants improved their hand localization accuracy when tested repetitively without performance feedback. Importantly, despite the reduction of average error, the spatial pattern of hand localization errors remained idiosyncratic. Based on individuals' hand localization performance, a standard convolutional neural network classifier could identify participants with good accuracy. Moreover, we did not find supporting evidence that participants' baseline hand localization performance could predict their motor performance in a visual Trajectory-matching task even though both tasks require accurate mapping of hand position to visual targets in the same workspace. Using a separate experiment, we not only replicated these findings but also ruled out the possibility that performance feedback during a few familiarization trials caused the observed improvement in hand localization. We conclude that the conventional hand localization test itself, even without feedback, can improve hand localization but leave the idiosyncrasy of hand localization map unchanged.
... Analyses on pre-test contrast thresholds for trained and untrained orientation across cue-validity condition and attention type show that participants were not at similar levels of performance prior to training. High inter-observer variability is commonly found in the field of PL [16,[85][86][87][88]. In fact, not only are individual differences in initial performance levels common in the field of PL but also the rate of learning is variable among participants. ...
The present study examined the role of exogenous and endogenous attention in task relevant visual perceptual learning (TR-VPL). VPL performance was assessed by examining the learning to a trained stimulus feature and transfer of learning to an untrained stimulus feature. To assess the differential role of attention in VPL, two types of attentional cues were manipulated; exogenous and endogenous. In order to assess the effectiveness of the attentional cue, the two types of attentional cues were further divided into three cue-validity conditions. Participants were trained, on a novel task, to detect the presence of a complex gabor patch embedded in fixed Gaussian contrast noise while contrast thresholds were varied. The results showed initial differences were found prior to training, and so the magnitude of learning was assessed. Exogenous and endogenous attention were both found to facilitate learning and feature transfer when investigating pre-test and post-test thresholds. However, examination of training data indicate attentional differences; with endogenous attention showing consistently lower contrast thresholds as compared to exogenous attention suggesting greater impact of training with endogenous attention. We conclude that several factors, including the use of stimuli that resulted in rapid learning, may have contributed to the generalization of learning found in the present study.
... Herein, we propose an optimization algorithm inspired by the learning mechanism of NC-crows developing Pandanus tools, called new NC-crow learning algorithm (NCCLA). Our motivations of introducing such algorithm are as follows: (i) the intelligent and distinguished behavior of learning mechanism of NC-crows to manufacture Pandanus tools; (ii) the previous studies highlighted that the self-improvement through the learning process is more direct and rapid than the natural evolution of genotypes [31,32]. As a result, employing the learning process techniques in NCcrows may lead to developing more effective algorithm than existing SI algorithms; (iii) it can also be noticed from the literature that the recently developed algorithms such as golden ratio optimization method (GROM) [33], the grey wolf optimizer (GWO) [34], the whale optimization algorithm (WOA) [35], the pathfinder algorithm (PFA) [36], the social learning algorithm (SLA) [16], and SLPSO [14] achieved better results and outperformed other traditional algorithms such as GA, ACO and PSO; (iv) the no free lunch (NFL) theorem states that the absolute superiority of an algorithm to solve all optimization algorithms cannot be claimed [37,38]. ...
Several metaheuristic algorithms have been introduced to solve different optimization problems. Such algorithms are inspired by a wide range of natural phenomena or behaviors. We introduced a new metaheuristic algorithm called âNew Caledonian (NC) crow learning algorithm (NCCLA),âù inspired by efficient social, asocial, and reinforcement mechanisms that NC-crows use to learn behaviors for developing tools from Pandanus trees to obtain food. Such mechanisms were modeled mathematically to develop NCCLA, whose performance was subsequently evaluated and statistically analyzed using 23 classical benchmark functions and 4 engineering problems. The results verify NCCLA’s performance efficiency and highlight its accelerated convergence and ability to escape from local minima. An extensive comparative study was conducted to demonstrate that the solution accuracy and convergence rate of NCCLA were better than those of other state-of-the-art metaheuristics. The results also indicate that NCCLA is a promising algorithm that can be applied to solve other optimization and real-world problems.
... The effects of consistent training on perceptual abilities have been extensively observed in structured experimental tasks. Indeed, laboratory studies reported many examples of dramatic improvements in sensitivity and speed after VPL training of primitive visual features, such as orientation (Schoups et al., 2001), motion (Ball and Sekuler, 1981) texture (Karni and Sagi, 1991) and Vernier acuity (Fahle and Edelman, 1993). In addition, it was demonstrated that such improvements can be highly specific to the trained stimulus features, for instance orientation (Schoups et al., 2001) and direction (Watanabe et al., 2002), although few other studies suggested that whether the learning effects are specific or generalizable depends on a number of factors such as task difficulty (Ahissar and Hochstein, 1993;Liu and Weinshall, 2000) and duration of the training (Jeter et al., 2010). ...
Visual system is endowed with an incredibly complex organization composed of multiple visual pathway affording both hierarchical and parallel processing. Even if most of the visual information is conveyed by the retina to the lateral geniculate nucleus of the thalamus and then to primary visual cortex, a wealth of alternative subcortical pathways is present. This complex organization is experience dependent and retains plastic properties throughout the lifespan enabling the system with a continuous update of its functions in response to variable external needs. Changes can be induced by several factors including learning and experience but can also be promoted by the use non-invasive brain stimulation techniques. Furthermore, besides the astonishing ability of our visual system to spontaneously reorganize after injuries, we now know that the exposure to specific rehabilitative training can produce not only important functional modifications but also long-lasting changes within cortical and subcortical structures.
The present review aims to update and address the current state of the art on these topics gathering studies that reported relevant modifications of visual functioning together with plastic changes within cortical and subcortical structures both in the healthy and in the lesioned visual system.
... During the last few decades, perceptual learning has been found to improve visual performance from simple visual feature discrimination to complex object recognition, such as contrast (Yu et al., 2004;Moret et al., 2018;Zhang et al., 2018), stimulus orientation (Wang et al., 2010;Jehee et al., 2012), motion (Ball and Sekuler, 1987;Larcombe et al., 2018), stereoacuity (Ding and Levi, 2011;Xi et al., 2014), vernier acuity (Fahle and Edelman, 1993;Hung and Seitz, 2014), shape (Gilbert and Sigman, 2000;Gilbert et al., 2009), and facial recognition (Gold et al., 1999). Previous studies have shown that the improvement in learning is highly specific to the characteristics of the trained task or stimulus (Fahle and Morgan, 1996;Li et al., 2004;Huang et al., 2007) or even to the trained eye or retinal location (Karni and Sagi, 1991). ...
It is well known that extensive practice of a perceptual task can improve visual performance, termed perceptual learning. The goal of the present study was to evaluate the dependency of visual improvements on the features of training stimuli (i.e., spatial frequency). Twenty-eight observers were divided into training and control groups. Visual acuity (VA) and contrast sensitivity function (CSF) were measured and compared before and after training. All observers in the training group were trained in a monocular grating detection task near their individual cutoff spatial frequencies. The results showed that perceptual learning induced significant visual improvement, which was dependent on the cutoff spatial frequency, with a greater improvement magnitude and transfer of perceptual learning observed for those trained with higher spatial frequencies. However, VA significantly improved following training but was not related to the cutoff spatial frequency. The results may broaden the understanding of the nature of the learning rule and the neural plasticity of different cortical areas.
... Over the course of the 6 training days, subjects trained for approximately 9 h and they completed 1,800 trials. In order to maximize learning efficiency, participants received feedback in the form of a short tone when they made a mistake (De Niear, Noel, & Wallace, 2017;Fahle & Edelman, 1993;Goldhacker, Rosengarth, Plank, & Greenlee, 2014). ...
Perceptual learning, the ability to improve the sensitivity of sensory perception through training, has been shown to exist in all sensory systems but the vestibular system. A previous study found no improvement of passive self-motion thresholds in the dark after intense direction discrimination training of either yaw rotations (stimulating semicircular canals) or y-translation (stimulating otoliths). The goal of the present study was to investigate whether perceptual learning of self-motion in the dark would occur when there is a simultaneous otolith and semicircular canal input, as is the case with roll tilt motion stimuli. Blindfolded subjects (n = 10) trained on a direction discrimination task with 0.2-Hz roll tilt motion stimuli (9 h of training, 1,800 trials). Before and after training, motion thresholds were measured in the dark for the trained motion and for three transfer conditions. We found that roll tilt sensitivity in the 0.2-Hz roll tilt condition was increased (i.e., thresholds decreased) after training but not for controls who were not exposed to training. This is the first demonstration of perceptual learning of passive self-motion direction discrimination in the dark. The results have potential therapeutic relevance as 0.2-Hz roll thresholds have been associated with poor performance on a clinical balance test that has been linked to more than a fivefold increase in falls.
... http://dx.doi.org/10.1101/850727 doi: bioRxiv preprint first posted online Nov. 21, 2019; even have found that the learning rate is similar with and without feedback in a 607 direction discrimination task (Fahle and Edelman 1993). These perceptual 608 improvements are generally attributed to the neural plasticity at the cellular level in 609 the visual system (Petrov et al. 2006). ...
Accumulating evidence indicates that the human's proprioception map appears subject-specific. However, whether the idiosyncratic pattern persists across time with good within-subject consistency has not been quantitatively examined. Here we measured the proprioception by a hand visual-matching task in multiple sessions over two days. We found that people improved their proprioception when tested repetitively without performance feedback. Importantly, despite the reduction of average error, the spatial pattern of proprioception errors remained idiosyncratic. Based on individuals' proprioceptive performance, a standard convolutional neural network classifier could identify people with good accuracy. We also found that subjects' baseline proprioceptive performance could not predict their motor performance in a visual trajectory-matching task even though both tasks require accurate mapping of hand position to visual targets in the same workspace. Using a separate experiment, we not only replicated these findings but also ruled out the possibility that performance feedback during a few familiarization trials caused the observed improvement in proprioception. We conclude that the conventional proprioception test itself, even without feedback, can improve proprioception but leave the idiosyncrasy of proprioception unchanged.
... The purpose of the first experiment was to establish the necessity of performance feedback when undertaking a period of training on a task to discriminate the direction of global motion. Fahle and Edelman (1993) predicted that internal reinforcement could act as the teaching signal when performance feedback was absent and when the confidence is high (Jeter et al., 2009;Talluri et al., 2015). Therefore, perceptual learning should occur when training procedures include a mixture of easy and difficult trials. ...
... Changes of the learning curve obtained under different experimental manipulations such as external noise (Lu, Chu, & Dosher, 2006;Lu & Dosher, 2004), training difficulty (J. Z. Liu & Weinshall, 2000), training schedule (Hung & Seitz, 2014;Xiao et al., 2008), feedback (Aberg & Herzog, 2012;Fahle & Edelman, 1993;J. Liu et al., 2010J. ...
The staircase method has been widely used in measuring perceptual learning. Recently, Zhao, Lesmes, and Lu (2017, 2019) developed the quick Change Detection (qCD) method and applied it to measure the trial-by-trial time course of dark adaptation. In the current study, we conducted two simulations to evaluate the performance of the 3-down/1-up staircase and qCD methods in measuring perceptual learning in a two-alternative forced-choice task. In Study 1, three observers with different time constants (40, 80, and 160 trials) of an exponential learning curve were simulated. Each simulated observer completed staircases with six step sizes (1%, 5%, 10%, 20%, 30%, and 60%) and a qCD procedure, each starting at five levels (+50%, +25%, 0, -25%, and -50% different from the true threshold in the first trial). We found the following results: Staircases with 1% and 5% step sizes failed to generate more than five reversals half of the time; and the bias and standard deviations of thresholds estimated from the post hoc segment-by-segment qCD analysis were much smaller than those from the staircase method with the other four step sizes. In Study 2, we simulated thresholds in the transfer phases with the same time constants and 50% transfer for each observer in Study 1. We found that the estimated transfer indexes from qCD showed smaller biases and standard deviations than those from the staircase method. In addition, rescoring the simulated data from the staircase method using the Bayesian estimation component of the qCD method resulted in much-improved estimates. We conclude that the qCD method characterizes the time course of perceptual learning and transfer more accurately, precisely, and efficiently than the staircase method, even with the optimal 10% step size.
... For example, training and improvements with vertical bisection stimuli ( Figure 1A) do not transfer to the same bisection stimulus rotated by 908 ( Figure 1B; e.g., Crist, Kapadia, Westheimer, & Gilbert, 1997;Crist, Li, & Gilbert, 2001;Grzeczkowski, Cretenoud, Herzog, & Mast, 2017;Grzeczkowski, Tartaglia, Mast, & Herzog, 2015;Herzog et al., 2012;Otto, Herzog, Fahle, Zhaoping, 2006;Parkosadze et al., 2008;Tartaglia, Balmert, Mast & Herzog, 2009;Tartaglia, Aberg, & Herzog, 2009). Perceptual learning was not only found to be specific for the trained stimulus orientation (Crist et al., 1997;Fahle & Edelman, 1993;Spang, Grimsen, Herzog, & Fahle, 2010;Vogels & Orban, 1985) but also for contrast (Sowden, Rose, & Davies, 2002;Yu, Klein, & Levi, 2004), motion direction (Ball & Sekuler, 1982, spatial frequency (Berardi & Fiorentini, 1987), retinal position (Ahissar & Hochstein, 1996), and even in some cases, the eye trained with (Karni & Sagi, 1991; but see Schoups & Orban, 1996). ...
Perceptual learning is usually feature-specific. Recently, we showed that perceptual learning is even specific for the motor response type. In a three-line bisection task, participants indicated whether the central line was offset either to the left or right by pressing a left or a right button, respectively. We found no transfer when the same participants adjusted the offset by using a computer mouse. Here, we first show that perceptual learning with mouse adjustments transfers to the untrained hand, but only for the trained adjustment condition. There was no transfer to the button press conditions, neither for the trained nor the untrained hand. Second, we show that a double training procedure enables transfer from the mouse adjustment to the button press condition. Hence, the specificity of perceptual learning to the motor response type can be overcome by double training as it is the case for visual features. Our results suggest that during perceptual learning, perceptuo-decisional signals are encoded together with the corresponding actions.
... J. Liu, Lu, & Dosher, 2012;Z. Liu & Weinshall, 2000), training schedule (Hung & Seitz, 2014;Xiao et al., 2008), feedback (Fahle & Edelman, 1993;Herzog & Fahle, 1997;J. Liu et al., 2012;Shibata, Yamagishi, Ishii, & Kawato, 2009), attention (Donovan, Szpiro, & Carrasco, 2015;Mukai et al., 2007;Szpiro, Lee, Wright, & Carrasco, 2013), and reward conditions (Seitz, Kim, & Watanabe, 2009;Zhang, Hou, et al., 2018) can reveal the impact of the manipulations on perceptual learning and inform and constrain computational models of perceptual learning Petrov, Dosher, & Lu, 2005). ...
The learning curve in perceptual learning is typically sampled in blocks of trials, which could result in imprecise and possibly biased estimates, especially when learning is rapid. Recently, Zhao, Lesmes, and Lu (2017, 2019) developed a Bayesian adaptive quick Change Detection (qCD) method to accurately, precisely, and efficiently assess the time course of perceptual sensitivity change. In this study, we implemented and tested the qCD method in assessing the learning curve in a four-alternative forced-choice global motion direction identification task in both simulations and a psychophysical experiment. The stimulus intensity in each trial was determined by the qCD, staircase or random stimulus selection (RSS) methods. Simulations showed that the accuracy (bias) and precision (standard deviation or confidence bounds) of the estimated learning curves from the qCD were much better than those obtained by the staircase and RSS method; this is true for both trial-by-trial and post hoc segment-by-segment qCD analyses. In the psychophysical experiment, the average half widths of the 68.2% credible interval of the estimated thresholds from the trial-by-trial and post hoc segment-by-segment qCD analyses were both quite small. Additionally, the overall estimates from the qCD and staircase methods matched extremely well in this task where the behavioral rate of learning is relatively slow. Our results suggest that the qCD method can precisely and accurately assess the trial-by-trial time course of perceptual learning.
... While the robustness of learning effects is well established, debate persists with respect to the mechanisms underlying VPL. Early psychophysical work found that learning effects are usually confined to the trained parameters 6,12 . Such strong specificity suggests that VPL most likely takes place within low-level visual areas (e.g., V1 or V2) since neurons therein exhibit narrow ranges of spatial and feature selectivity (e.g., orientation, motion direction). ...
Visual perceptual learning (VPL) can lead to long-lasting perceptual improvements. One of the central topics in VPL studies is the locus of plasticity in the visual processing hierarchy. Here, we tackled this question in the context of motion processing. We took advantage of an established transition from component-dependent representations at the earliest level to pattern-dependent representations at the middle-level of cortical motion processing. Two groups of participants were trained on the same motion direction identification task using either grating or plaid stimuli. A set of pre- and post-training tests was used to determine the degree of learning specificity and generalizability. This approach allowed us to disentangle contributions from different levels of processing stages to behavioral improvements. We observed a complete bi-directional transfer of learning between component and pattern stimuli that moved to the same directions, indicating learning-induced plasticity associated with intermediate levels of motion processing. Moreover, we found that motion VPL is specific to the trained stimulus direction, speed, size, and contrast, diminishing the possibility of non-sensory decision-level enhancements. Taken together, these results indicate that, at least for the type of stimuli and the task used here, motion VPL most likely alters visual computation associated with signals at the middle stage of motion processing.
This book was the first handbook where the world's foremost 'experts on expertise' reviewed our scientific knowledge on expertise and expert performance and how experts may differ from non-experts in terms of their development, training, reasoning, knowledge, social support, and innate talent. Methods are described for the study of experts' knowledge and their performance of representative tasks from their domain of expertise. The development of expertise is also studied by retrospective interviews and the daily lives of experts are studied with diaries. In 15 major domains of expertise, the leading researchers summarize our knowledge on the structure and acquisition of expert skill and knowledge and discuss future prospects. General issues that cut across most domains are reviewed in chapters on various aspects of expertise such as general and practical intelligence, differences in brain activity, self-regulated learning, deliberate practice, aging, knowledge management, and creativity.
Perceptual learning (PL) involves long-lasting improvement in perceptual tasks following extensive training and is accompanied by modified neuronal responses in sensory cortical areas in the brain. Understanding the dynamics of PL and the resultant synaptic changes is important for causally connecting PL to the observed neural plasticity. This is theoretically challenging because learning-related changes are distributed across many stages of the sensory hierarchy. In this paper, we modeled the sensory hierarchy as a deep nonlinear neural network and studied PL of fine discrimination, a common and well-studied paradigm of PL. Using tools from statistical physics, we developed a mean-field theory of the network in the limit of a large number of neurons and large number of examples. Our theory suggests that, in this thermodynamic limit, the input-output function of the network can be exactly mapped to that of a deep linear network, allowing us to characterize the space of solutions for the task. Surprisingly, we found that modifying synaptic weights in the first layer of the hierarchy is both sufficient and necessary for PL. To address the degeneracy of the space of solutions, we postulate that PL dynamics are constrained by a normative minimum perturbation (MP) principle, which favors weight matrices with minimal changes relative to their prelearning values. Interestingly, MP plasticity induces changes to weights and neural representations in all layers of the network, except for the readout weight vector. While weight changes in higher layers are not necessary for learning, they help reduce overall perturbation to the network. In addition, such plasticity can be learned simply through slow learning. We further elucidate the properties of MP changes and compare them against experimental findings. Overall, our statistical mechanics theory of PL provides mechanistic and normative understanding of several important empirical findings of PL.
It is widely believed that feedback improves behavior, but the mechanisms behind this improvement remain unclear. Different theories postulate that feedback has either a direct effect on performance through automatic reinforcement mechanisms or only an indirect effect mediated by a deliberate change in strategy. To adjudicate between these competing accounts, we performed two large experiments on human adults (total N = 518); approximately half the participants received trial-by-trial feedback on a perceptual task, whereas the other half did not receive any feedback. We found that feedback had no effect on either perceptual or metacognitive sensitivity even after 7 days of training. On the other hand, feedback significantly affected participants' response strategies by reducing response bias and improving confidence calibration. These results suggest that the beneficial effects of feedback stem from allowing people to adjust their strategies for performing the task and not from direct reinforcement mechanisms, at least in the domain of perception.
Training to optimize visual performance abilities is a logical supplement to other training programs that athletes perform in order to improve sports performance in competition. An overview of how to develop and successfully implement sports vision training is presented. A detailed description of common sports vision training approaches is provided with research evidence to support efficacy, when available. Updates on commercially available instrumentation to train various visual performance abilities is presented.
Visual perceptual learning (VPL) can lead to long-lasting perceptual improvements. While the efficacy of VPL is well established, there is still a considerable debate about what mechanisms underlie the effects of VPL. Much of this debate concentrates on where along the visual processing hierarchy behaviorally relevant plasticity takes place. Here, we aimed to tackle this question in context of motion processing, a domain where links between behavior and processing hierarchy are well established. Specifically, we took advantage of an established transition from component-dependent representations at the earliest level to pattern-dependent representations at the middle-level of cortical motion processing. We trained two groups of participants on the same motion direction identification task using either grating or plaid stimuli. A set of pre- and post-training tests was used to determine the degree of learning specificity and generalizability. This approach allowed us to disentangle contributions from both low- and mid-level motion processing, as well as high-level cognitive changes. We observed a complete bi-directional transfer of learning between component and pattern stimuli as long as they shared the same apparent motion direction. This result indicates learning-induced plasticity at intermediate levels of motion processing. Moreover, we found that motion VPL is specific to the trained stimulus direction, speed, size, and contrast, highlighting the pivotal role of basic visual features in VPL, and diminishing the possibility of non-sensory decision-level enhancements. Taken together, our study psychophysically examined a variety of factors mediating motion VPL, and demonstrated that motion VPL most likely alters visual computation in the middle stage of motion processing.
There have been long-standing debates regarding whether supervised or unsupervised learning mechanisms are involved in visual perceptual learning (VPL) [1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14]. However, these debates have been based on the effects of simple feedback only about response accuracy in detection or discrimination tasks of low-level visual features such as orientation [15, 16, 17, 18, 19, 20, 21, 22]. Here, we examined whether the content of response feedback plays a critical role for the acquisition and long-term retention of VPL of complex natural images. We trained three groups of human subjects (n = 72 in total) to better detect “grouped microcalcifications” or “architectural distortion” lesions (referred to as calcification and distortion in the following) in mammograms either with no trial-by-trial feedback, partial trial-by-trial feedback (response correctness only), or detailed trial-by-trial feedback (response correctness and target location). Distortion lesions consist of more complex visual structures than calcification lesions [23, 24, 25, 26]. We found that partial feedback is necessary for VPL of calcifications, whereas detailed feedback is required for VPL of distortions. Furthermore, detailed feedback during training is necessary for VPL of distortion and calcification lesions to be retained for 6 months. These results show that although supervised learning is heavily involved in VPL of complex natural images, the extent of supervision for VPL varies across different types of complex natural images. Such differential requirements for VPL to improve the detectability of lesions in mammograms are potentially informative for the professional training of radiologists.
Perceptual learning is typically highly specific to the stimuli and task used during training. However, recently, it has been shown that training on global motion can transfer to untrained tasks, reflecting the generalising properties of mechanisms at this level of processing. We investigated (i) if feedback was required for learning in a motion coherence task, (ii) the transfer across the spatial frequency of training on a global motion coherence task and (iii) the transfer of this training to a measure of contrast sensitivity. For our first experiment, two groups, with and without feedback, trained for ten days on a broadband motion coherence task. Results indicated that feedback was a requirement for robust learning. For the second experiment, training consisted of five days of direction discrimination using one of three motion coherence stimuli (where individual elements were comprised of either broadband Gaussian blobs or low- or high-frequency random-dot Gabor patches), with trial-by-trial auditory feedback. A pre- and post-training assessment was conducted for each of the three types of global motion coherence conditions and high and low spatial frequency contrast sensitivity (both without feedback). Our training paradigm was successful at eliciting improvement in the trained tasks over the five days. Post-training assessments found evidence of transfer for the motion coherence task exclusively for the group trained on low spatial frequency elements. For the contrast sensitivity tasks, improved performance was observed for low- and high-frequency stimuli, following motion coherence training with broadband stimuli, and for low-frequency stimuli, following low-frequency training. Our findings are consistent with perceptual learning, which depends on the global stage of motion processing in higher cortical areas, which is broadly tuned for spatial frequency, with a preference for low frequencies.
The ability to detect small differences in the positions of two lines (vernier acuity) showed some improvement with practice
in all eight subjects, even for subjects given no error feedback. The average decline in threshold with training (2,000–2,500
responses) was about 40%. We used three target orientations: vertical, horizontal, and right oblique. Orientational differences
remained stable in only one subject. In five subjects, orientational differences present at the beginning of training diminished
or disappeared with increased experience; in two, they increased.
In many different spatial discrimination tasks, such as in determining the sign of the offset in a vernier stimulus, the human
visual system exhibits hyperacuity by evaluating spatial relations with the precision of a fraction of a photoreceptor's diameter.
It is proposed that this impressive performance depends in part on a fast learning process that uses relatively few examples
and that occurs at an early processing stage in the visual pathway. This hypothesis is given support by the demonstration
that it is possible to synthesize, from a small number of examples of a given task, a simple network that attains the required
performance level. Psychophysical experiments agree with some of the key predictions of the model. In particular, fast stimulus-specific
learning is found to take place in the human visual system, and this learning does not transfer between two slightly different
hyperacuity tasks.
The most frequent cause of visual loss in childhood is functional amblyopia, an abnormality of visual acuity usually associated with either anisometropia (unequal refractive errors) or strabismus (turned eye) during early development. The usual clinical investigation of the visual acuity of amblyopes involves discrimination of the high contrast letters of a Snellen chart; however, there are other aspects of acuity, for example, grating acuity (the high spatial frequency limit of vision) and Vernier acuity (the smaller perceptible misalignment). Because of the extreme precision of Vernier acuity compared with either grafting or Snellen acuity, it is considered to be a form of hyperacuity which requires very precise positional information. In an effort to understand the nature of the neural abnormalities which cause the reduced acuity of amblyopes, we have measured here the Vernier acuity of amblyopic observers using an extended Vernier grating stimulus, and compared these results with their Snellen acuity and grating acuity. The results showed that different acuity losses are associated wih anisometropic versus strabismic amblyopia. When scaled with respect to their grating acuity, anisometropic amblyopes, like normals, showed hyperacuity, even at high spatial frequencies, while strabismic amblyopes showed severe losses in Vernier acuity. Snellen letter acuity showed a similar deficit relative to grating acuity in strabismic but not in anisometropic amblyopes. Contrary to some previous theories which have considered that all forms of amblyopia share a common neural basis, these results strongly support the view that different neural losses are associated with amblyopias of different aetiologies.
"Adaptation to disarranged hand-eye coordination is claimed to occur only during prolonged exposure involving visual stimulation dependent upon S's self-produced arm movements. Sensory stimulation that is dependent upon self-produced movements is called re-afference by v. Holst to distinguish it from ex-afference or sensory stimulation not dependent upon S's movements. Reductions of errors in localization—induced by prisms placed in front of the eyes—were compared after controlled exposures with and without re-afferent visual stimulation from the moving hand. The results [of 6 and 8 Ss] clearly indicate that re-afferent stimulation is necessary for adaptation." (PsycINFO Database Record (c) 2012 APA, all rights reserved)
Networks that solve specific visual tasks, such as the evaluation of spatial relations with hyperacuity precision, can be easily synthesized from a small set of examples. The present paper describes a series of simulated psychophysical experiments that replicate human performance in hyperacuity tasks. The experiments were conducted with a detailed computational model of perceptual learning, based on HyperBF interpolation. The success of the simulations provides a new angle on the purposive aspect of human vision, in which the capability for solving any given task emerges only if the need for it is dictated by the environment. We conjecture that almost any tractable psychophysical task can be performed better after suitable training, provided the necessary information is available in the stimulus.
Line orientation discrimination improves with selective practice for oblique orientations and not for principal orientations. This training effect was observed with an identification task as well as with two alternative forced choice tasks. Despite the improvement for oblique orientations, just noticeable differences in orientation are still larger for the practised oblique orientation than for the principal orientations after 5000 practice trials. These findings suggest that the oblique effect in line orientation has at least two sensorial components, one of which is attributed to the meridional variations in the preferred orientation of area 17 S-cells.
The ability of human observers to discriminate the orientation of a pair of straight lines differing by 3 degrees improved with practice. The improvement did not transfer across hemifield or across quadrants within the same hemifield. The practice effect occurred whether or not observers were given feedback. However, orientation discrimination did not improve when observers attended to brightness rather than orientation of the lines. This suggests that cognitive set affects tuning in retinally local orientation channels (perhaps by guiding some form of unsupervised learning mechanism) and that retinotopic feature extraction may not be wholly preattentive.
The effect of training on an observer's ability to detect the misalignment of three points, a hyperacuity, and to resolve a six-line grating was studied in a transfer-of-training design with observers (4 in each of two experiments) who were experienced in making psychophysical judgments of other visual stimuli. The transfer-of-training design enabled us to look for any training-based improvement. Long periods of training produced no statistically significant improvement in performance under any condition. There were small practice-based improvements, but the primary patterns indicated threshold fluctuation rather than improvement. We interpret the results to indicate that the neural mechanisms underlying three-point alignment and grating discrimination, like those for gap bisection (Klein & Levi, 1985), are not malleable to any significant extent.
If the two segments of a vernier target are presented to different eyes, i.e. dichoptically, thresholds are three to four times higher than with presentation to the same eye. This increase in thresholds is mainly due to uncorrelated movements of both eyes, such as tremor and drifts, that occur even under steady fixation. The psychophysically measured thresholds allow one to calculate an upper estimate for the amplitudes of uncorrelated eye movements during fixation. This estimate matches the best results from direct eye position recording, with the calculated mean amplitude of eye tremor corresponding to roughly one photoreceptor diameter. The combined amplitude of both correlated and uncorrelated eye movements was also measured by delaying one segment of the vernier relative to its partner under monocular or dichoptic conditions. Fixation proved to be relatively stable, and trained observers could sustain eye position within a few min arc.
In terms of functional anatomy, where does learning occur when, for a basic visual discrimination task, performance improves with practice (perceptual learning)? We report remarkable long-term learning in a simple texture discrimination task where learning is specific for retinal input. This learning is (i) local (in a retinotopic sense), (ii) orientation specific but asymmetric (it is specific for background but not for target-element orientation), and (iii) strongly monocular (there is little interocular transfer of learning). Our results suggest that learning involves experience-dependent changes at a level of the visual system where monocularity and the retinotopic organization of the visual input are still retained and where different orientations are processed separately. These results can be interpreted in terms of local plasticity induced by retinal input in early visual processing in human adults, presumably at the level of orientation-gradient sensitive cells in primary visual cortex.
With training, an observer's ability to discriminate similar directions of motion gradually improves. A series of studies reveals that this improvement, (1) is restricted to the trained direction and other, similar directions, (2) persists for at least several months, (4) shows appreciable, but not complete, transfer between the two eyes, and (5) is largely restricted to the stimulated region of the field. Moreover, the improvement in direction discrimination does not produce a concomitant change in detection thresholds. In all likelihood, most of the improvement in direction discrimination represents a change in visual function, rather than changes in nonsensory processes.
Anisometropic amblyopes were found to have a reduced sensitivity for shape discrimination. The introduction of positional jitter in the elements of the display had a profound effect on the performance of the normal eye, but not on that of the amblyopic eye. On the other hand the introduction of gaussian blur affects the performance of both eyes to the same degree. We conclude that raised spatial uncertainty due to metrical scrambling is a suitable model for anisometropic amblyopia.
When a vernier target is flanked by a pair of optimally positioned flanks, offset discrimination is strongly degraded. Spatial interference with vernier acuity was studied in each eye of observers with unilateral amblyopia associated with strabismus, anisometropia or both, and were compared to the functions obtained in the normal periphery (Levi et al., 1985). The results showed that: (1) For both strabismic and anisometropic amblyopes, as in normal central and peripheral vision, the extent of spatial interference was proportional to the unflanked vernier threshold. (2) For anisometropic amblyopes, grating and vernier acuity are affected similarly. (3) For strabismic amblyopes, like the normal periphery, vernier and grating acuity are decoupled, with vernier falling off faster than grating acuity. (4) The preferred eyes of strabismic but not anisometropic amblyopes have poorer vernier acuity than the normal controls. A conceptual framework for amblyopia based upon spatial filtering and spatial sampling is discussed.
The specification of intensity of self-luminous point and line targets, such as those created on cathode-ray tubes, is discussed and methods are outlined for measuring it and deducing the resulting retinal illuminance. The calculations can be extended to the specification of contrast by combining them with the more traditional method of identifying the retinal illuminance provided by uniform backgrounds. Such values of contrast are not independent of observation distance.
Inexperienced observers show a delay before experiencing the stereoscopic percept from a random dot stereo pair. This perception time is progressively reduced with repeated exposures of the stereogram. The authors have investigated the specificity of this perceptual learning effect, using stereograms made up of short oblique line elements. Learning with a stereogram consisting of 45° line elements transferred completely to an uncorrelated pattern with the same element orientation, but there was a marked failure of transfer to a pattern whose elements had the opposed oblique orientation. Thus the stereoscopic skill that has been acquired may be specific to those orientation analyzers that were stimulated during the training period.
Stimuli on corresponding points of both retinae that cannot be fused may cause binocular rivalry: the stimuli suppress each other alternately. This effect was used to study the influence of image sharpness upon binocular inhibition. Blurring an image means decreasing its contrast and attenuating its high spatial frequencies. Both factors diminish the time that a stimulus is perceived during rivalry. This fact has implications both for normal vision--as objects off the horopter are normally blurred--and for disturbed vision when the image of one or both eyes is (locally) deteriorated. In both cases, the binocular field of view can be combined from the 'good' parts of both eyes. Hence, the field of view may consist, in a piece-meal fashion, of parts stemming from the right or the left eye exclusively and others where both images are superimposed. We present evidence for the hypothesis that there is a common neural mechanism causing both binocular rivalry and functional amblyopia in anisometropia and strabismus. Consequences of the results on rivalry suppression for the pathophysiology and therapy of strabismic amblyopia are discussed.
Spatial localization was investigated for each eye of amblyopic observers using a bisection paradigm. The stimuli were comprised of a grating composed of bright lines, and a test line. The test line was either placed above the grating (bisection-no overlap) or within the row of lines comprising the grating (bisection-with overlap) and thresholds for each bisection task were measured as a function of the fundamental spatial frequency of the grating. Vernier thresholds were also measured. For the nonamblyopic eyes at low spatial frequencies, bisection thresholds were a constant fraction ("Weber" fraction) of the space to be bisected, while at high spatial frequencies thresholds were approximately a constant retinal distance (a hyperacuity). However the spatial localization of an amblyopic eye depends upon both the type of amblyopia, and the stimulus configuration. Specifically, for anisometropic amblyopia, spatial localization (bisection-no overlap) and vernier, when scaled to the resolution losses, were normal. However, spatial adjacency (bisection with overlap), while enhancing the spatial localization of nonamblyopic eyes at high spatial frequencies, markedly elevated thresholds in the amblyopic eyes of anisometropic amblyopes. Strabismic amblyopes on the other hand show disturbances in both spatial localization tasks which can not be accounted for on the basis of reduced resolution. Their results are characterized by an absence of a constant Weber fraction at low spatial frequencies and "crowding" effects at high spatial frequencies. For strabismic amblyopes, the optimal localization thresholds were similar to the Snellen threshold, while for anisometropic amblyopes, the optimal localization thresholds were several times better than the Snellen threshold.
The improvement in stereoacuity of two inexperienced, normal subjects was compared at foveal and at 2.5 degrees and 5 degrees peripheral target locations as a function of practice. Outlines of two squares differing only in binocular disparity were used as test stimuli and estimates of stereoacuity were obtained by application of the method of constant stimuli with feedback. The peripheral thresholds of both subjects improved 60-80% over the course of the first 3000-4000 responses at each stimulus location. Foveal improvement followed an identical time-course with a 73% improvement in one subject and only 23% in the other. This difference was reflected in the peripheral/foveal threshold ratios of the two subjects and underlines the necessity of ensuring the stability of thresholds. Stereoacuity measurements were also obtained using several different square separations at the fovea and at 2.5 degrees, 5 degrees and 10 degrees peripheral locations along the horizontal and vertical retinal meridians of two other normal subjects. Practice-stabilized disparity thresholds using optimal target separations revealed a steeper deterioration between the fovea and 2.5-5 degrees eccentricities than did measurements of the same subjects' minimum angles of resolution (MAR). The decrease of optimal stereoacuity at the more peripheral test locations was more gradual than has been previously reported but was not clearly related to that of the MAR.
The effects of practice in the discrimination of briefly flashed gratings were investigated by a forced-choice procedure with error correction in a number of tasks requiring discimination either of pairs of complex gratings of different waveforms or of “simple” (sinusoidal) gratings of slightly different spatial frequency. The percentage of correct responses progressively increases with repetition of trials up to 100–200 trials and then levels off, remaining rather constant thereafter even after days or weeks, in all tasks involving discrimination of complex gratings.However, when the gratings are set perpendicular to those used for the training sessions, or their spatial frequency is changed by 1 octave, the effects of previous perceptual learning are lost, while transfer of learning effects is obtained for smaller changes in orientation (±30°) or spatial frequency (±12 octave). The spatial frequency discrimination of sinusoidal gratings does not improve with a comparable number of trials.
Several examples of 'perceptual learning' (improvement of some perceptual task with practice) have been reported. These studies are of great interest for neurological research because they demonstrate plasticity of the nervous system. Even for apparently basic perceptual tasks, such as visual acuity or vernier acuity, practice can facilitate a neural change which enhances performance. One question in this field is where does this learning occur? Indications about the possible neural site of a learning process may be derived from its specificity for some particular stimulus parameters. For instance, there is a hint that learning in global stereopsis may occur at a stage where visual information is processed by mechanisms selectively sensitive to different stimulus orientations. We report here an experiment on perceptual learning in the discrimination of gratings of different waveform. Our findings show that learning is specific for both the orientation and the spatial frequency of the practice stimulus.
Learning an input-output mapping from a set of examples, of the type that many neural networks have been constructed to perform,
can be regarded as synthesizing an approximation of a multidimensional function (that is, solving the problem of hypersurface
reconstruction). From this point of view, this form of learning is closely related to classical approximation techniques,
such as generalized splines and regularization theory. A theory is reported that shows the equivalence between regularization
and a class of three-layer networks called regularization networks or hyper basis functions. These networks are not only equivalent
to generalized splines but are also closely related to the classical radial basis functions used for interpolation tasks and
to several pattern recognition and neural network algorithms. They also have an interesting interpretation in terms of prototypes
that are synthesized and optimally combined during the learning stage.
Hyperacuity IO, vision and visual dysfunction (Chap. 4) Binocular vision and ocular motility
- M J Morgan
Morgan, M. J. (1991). Hyperacuity. In Regan, D. (Ed.), Spatial vision, Vol. IO, vision and visual dysfunction (Chap. 4). London: Macmillan. von Noorden, G. K. (1990). Binocular vision and ocular motility. St Louis, MO.: Mosby.
Synthesis of visual modules from examples: Learning hyperacuity.In Artificial intelli-gence memo 1271 Fast perceptual learning in visual hyperacuity
- T Poggio
- M Fahle
- S Edelman
Poggio, T., Fahle, M. & Edelman, S. (1991). Synthesis of visual modules from examples: Learning hyperacuity.In Artificial intelli-gence memo 1271. Cambridge, Mass.: MIT Press, Poggio, T., Fahle, M. & Edelman, S. (1992b). Fast perceptual learning in visual hyperacuity. Science, 256, 1018-1021.
The effect of practice on the oblique effect in line orientation judgments
- Vogels