Hand Position Alters Vision by Modulating the Time Course of Spatial Frequency Use

Abstract

The nervous system gives preferential treatment to objects near the hands that are candidates for action. It is not yet understood how this process is achieved. Here we show evidence for the mechanism that underlies this process having used an experimental technique that maps the use of spatial frequencies (SFs) during object recognition across time. We used this technique to replicate and characterize with greater precision the coarse-to-fine SF sampling observed in previous studies. Then we show that the visual processing of real-world objects near an observer's hands is biased toward the use of low-SF information, around 288 ms. Conversely, high-SF information presented around 113 ms impaired object recognition when objects were presented near the hands. Notably, both of these effects happened relatively late during object recognition and suggest that the modulation of SF use by hand position is at least partly attentional in nature.

Full text

Hand Position & Spatial Frequency Use
1
Hand position alters vision by modulating the time course of spatial frequency use
Laurent Caplette1, Bruno Wicker2, Frédéric Gosselin1 & Greg L. West1
1CERNEC, Département de psychologie, Université de Montréal
2Aix Marseille Univ, CNRS, LNC, Marseille, France.
Address for Correspondence:
Dr. Greg L. West
University of Montreal
Department of Psychology
Pavillon Marie-Victorin
90, avenue Vincent d'Indy
Montreal QC
H2V 2S9
email: gregory.west@umontreal.ca

Hand Position & Spatial Frequency Use
2
Abstract
The nervous system gives preferential treatment to objects near the hands that are candidates for
action. It is not yet understood how this process is achieved. Here we show evidence for the
mechanism that underlies this process using an experimental technique that maps the use of
spatial frequencies (SFs) across time during object recognition. First, we use this technique to
replicate and characterize with greater precision the coarse-to-fine SF sampling observed in
previous studies. Then, we show that the visual processing of real world objects near an
observer’s hands is biased towards the use of low SF information around 288 ms. Conversely,
high SFs presented around 113 ms impaired object recognition when objects were presented near
the hands. Importantly, both these effects happen relatively late during object recognition and
suggest that the modulation of SF use by hand position is at least partly attentional in nature.
Key Words: action-perception; spatial frequencies; magnocellular pathway; hand-position;
embodied cognition; object recognition.

Hand Position & Spatial Frequency Use
3
Introduction
Human perception and action systems interact to produce very accurate visually guided
movements to accomplish everyday tasks (e.g., reaching and grasping). It is therefore not
surprising that action can have a large effect on perceptual processes. Research during the last
decade has demonstrated that performance is affected by the type of action being performed and
the spatial relationship between an observer’s effectors (e.g., hands, tools) and the target object.
More recently, it has been hypothesized that the effect of hand proximity on vision
represents a biasing of visual processing towards pathways responsible for different aspects of
visual input (i.e., perception and action). Current models of vision propose that visual processing
is divided into two major pathways known as the parvocellular (P) and magnocellular (M)
systems, whose separation begins at the retinal level, and is responsible for the functional
distinction between visual perception and vision for action. Further, the ventral-perception visual
stream has a larger number of projections from the P pathway while the dorsal-action visual
stream has a larger number of projections from the M pathway.
Crucially, the M and P pathways preferentially treat separate bands of spatial frequencies
(SF): low SFs, which provide coarse visual information, are extracted early and processed
through the fast acting M pathway, while conversely, high SFs, which provide finer visual
information, are extracted later and processed more slowly by the P pathway . This coarse-to-fine
SF extraction has been observed behaviorally in numerous studies (e.g., Hughes, Nozawa &
Kitterle, 1996; Hupe et al., 2001; Schyns & Oliva, 1994; Caplette et al., 2016). Further, this
temporal distinction between the processing of low and high SFs is present in both the early
visual cortex (Goddard et al., 2016; Jemel et al., 2010; Parker & Salzen, 1977) and the frontal
cortex (Bar et al., 2006; Goddard et al., 2016).

Hand Position & Spatial Frequency Use
4
A growing body of evidence suggests that hand position near a stimulus can bias visual
processing toward the action-oriented M pathway that preferentially treats low SFs and impair the
processing of high SF information conducted along the perception-oriented P pathway. This
mechanism is hypothesized to facilitate precise interaction with objects that are candidates for
action by up-signaling visual information conducted along the dorsal-action M pathway and
down-signaling perceptual information along the P pathway.
At this point, many aspects of the effect of hands on SF sampling remain unclear: (a) what
specific SF bands during visual processing are differentially affected by hand position, (b) what
is the impact of hand position on the visual treatment of ecologically valid objects that people
would find in their everyday lives, and importantly (c) at which stage or stages of object
recognition does hand position affect SF sampling? To address these research questions, we
employed a technique that maps the use of SFs contained in everyday objects across time with
unprecedented resolution. More specifically, we created dynamic stimuli from still images (a
bench, a pale, a plant, a wrapped gift, a cake, etc.), which were presented as 333 ms videos that
randomly revealed SF bands (ranging from 0.5 to 128 cycles per image, cpi) at variable time
points (ranging from early to late time points within the video). In Experiment 1, we tested the
value and the reliability of this method by examining the time course of SF sampling during
object recognition with hands in a typical downward position. We expected to find the coarse-to-
fine sampling that has been observed in past studies (e.g., Caplette et al., 2016; Hughes et al.,
1996; Hupe et al., 2001; Schyns & Oliva, 1994). In a second experiment, with a new set of
subjects, we investigated the impact of hand position on this pattern of SF extraction by
contrasting conditions where subjects placed their hands either near or far from the stimulus.
General Method

Hand Position & Spatial Frequency Use
5
Materials
The experimental programs ran on Mac Pro (Apple Inc.) computers in the Matlab
(Mathworks Inc.) environment, using functions from the Psychophysics Toolbox (Brainard,
1997; Pelli, 1997). All stimuli were presented on Asus VG278H monitors (1920 x 1080 pixels at
120 Hz), calibrated to allow linear manipulation of luminance. Luminance ranged from 1.6 cd/m2
to 159 cd/m2.
Stimuli
Eighty-six object grayscale images of everyday man-made objects were selected from the
database used in Shenhav et al. (2013) and from Internet searches. Images were 256 x 256 pixels
and median object width was 220 pixels. The objects were cropped manually and pasted on a
homogenous mid-gray background. The spatial frequency (SF) spectrum of each image was set to
the mean SF spectrum of the images and mean luminance was equalized across images using the
SHINE toolbox (Willenbockel et al., 2010). Resulting images had a root mean square (RMS)
contrast of about 0.20.
On each trial, participants were shown a short video (333 ms) consisting of an object
image with random SFs gradually revealed at random time points (e.g., Video S1; Video S2); that
is, on each video frame, there would typically be several SFs shown among all possible SFs, and
these would change from frame to frame. To create these dynamic stimuli, we first randomly
generated, on each trial, a matrix of dimensions 256 x 40 (representing respectively SFs from 0.5
to 128 cpi, and frames, each lasting 8.33 ms) in which most elements were zeros and a few were
ones. The number of ones was adjusted on a trial-by-trial basis to maintain performance at 75%
correct. We then convolved this sparse matrix with a 2D Gaussian kernel (a “bubble”; σSF = 1.5
cpi; σtime = 15 ms). This resulted in the trial’s sampling matrix: a SF x time plane with randomly
located bubbles. Every column of this sampling matrix was then rotated around its origin to

Hand Position & Spatial Frequency Use
6
create isotropic 2D random filters. Finally, these 2D random filters were dot-multiplied by the
base image's spectrum and inverse fast Fourier transformed to create a filtered version of the
image for every video frame (see Figure 1 for an illustration of this method). To ensure accurate
luminance display, we applied noisy-bit dithering to the final stimuli (Allard & Faubert, 2008).
Procedure
Participants sat in front of a computer monitor, in a dim-lighted room. They completed
two 500-trial blocks on the first day and two more on a second day. A short break occurred every
50 trials. Each trial was comprised of the following events: a fixation cross (300 ms), a blank
screen (200 ms), the video stimulus (333 ms), a fixation cross (300 ms), a blank screen (200 ms),
and an object name at the basic level of abstraction that remained on screen either until a
response was provided or for a maximum of 1 s, in which case it was replaced by a blank screen
until a response was provided. The number of bubbles was adjusted on a trial-by-trial basis using
a gradient descent algorithm to maintain performance at 75% correct. Subjects were asked to
indicate whether the name matched the object as accurately and as rapidly as possible. The basic-
level name and the object matched 50% of the time; on the trials in which they didn't match, the
name was randomly chosen among the basic-level names of all other objects.
Regression analysis
Accuracies and response times were transformed into z-scores for every object (separately
for each condition in experiment 2) to minimize variability due to differences in object
recognizability or familiarity with the object name. Further, z-scores were calculated for each
500-trial block to diminish variability due to task learning, and for each subject to minimize
residual individual differences in performance. Trials associated with z-scores over 3 or below -3
(either in accuracy or response times) were discarded from the regressions (2.23% of trials in
experiment 1; 0.26% of trials in experiment 2).

Hand Position & Spatial Frequency Use
7
To uncover which spatial frequencies in which time frames led to accurate object
recognition, we performed multiple least-square linear regressions between accuracies and
corresponding sparse matrices, separately for each subject (and each condition, in experiment 2).
The resulting matrices of regression coefficients were then summed across subjects and
convolved with a Gaussian kernel (σSF = 5 cpi; σtime = 42 ms) — henceforth we shall refer to
these matrices as classification images. The same procedure was repeated with 500 bootstrapped
samples, which were then used to transform the summed regression coefficients into z-scores.
Finally, we applied a Cluster test (Chauvin, Worsley, Schyns, Arguin, & Gosselin, 2005) to the
classification images to assess their statistical significance. Given an arbitrary z-score threshold
(here ±3.5), this test gives a cluster size k above which the specified p (here .05, two-tailed) is
satisfied, controlling the Family-Wise Error Rate (FWER) while taking into account the
correlation in the data.
Experiment 1
In Experiment 1, we tested the value of this new method by examining the time course of
SF sampling during object recognition with hands in a typical downward position. We expected
to find the classic coarse-to-fine sampling that has been observed in past studies (e.g., Caplette et
al., 2016; Hughes et al., 1996; Hupe et al., 2001; Schyns & Oliva, 1994).
Method
Twenty-three right-handed adult participants (10 males; mean age = 22.14; SD = 1.85)
were recruited on the campus of the University of Montreal. Subjects had normal or corrected to
normal vision, and did not suffer from any visual or reading disability. The study was approved
by the ethics board of the University of Montreal's Faculty of Arts and Sciences. Written consent
from all participants was obtained after the procedure had been fully explained, and a monetary

Hand Position & Spatial Frequency Use
8
compensation was provided upon completion of the experiment. During the task, chin rests were
used to maintain viewing distance at 76 cm; images subtended 6 x 6 degrees of visual angle.
Results and discussion
Participants responded correctly on an average of 75.02% of the trials and required an
average of 84.32 bubbles to do so. The mean response time was 719 ms. The z-scored group
classification image is illustrated in Figure 2. We included the SFs from 0.08 to 9.83 cycles per
degree (cpd; equivalent in this experiment to 0.5 to 59 cpi) in our analyses, because they have
been consistently identified as contributing to accurate object recognition (e.g., Caplette et al.,
2014, 2016; Gold et al., 1999). The z-scores indicate the correlation between the presentation of a
given SF on a given time frame and accuracy; white curves indicate significant clusters (p < .05,
two-tailed, FWER-corrected). This analysis revealed a first earlier significant cluster that peaked
at 2.25 cpd and 13 ms (Zmax = 4.49, k =148) and led to accurate object recognition. A second later
significant cluster peaking at 5.08 cpd and 304 ms (Zmax = 5.05, k = 1240) also led to accurate
object recognition.
To reduce the dimensionality of the results and characterize them more concisely, we
fitted a linear model on the classification image. The model consisted of a surface defined by the
inequalities
a1+b1t<f<a2+b2t
, where f stands for spatial frequency (cpd), t for time (s), and
a1, a2, b1 and b2 are free parameters. The model was fitted using the Nelder-Mead simplex
method. The best fitting model (R2 = 0.67) displays a clear coarse-to-fine pattern, in which the
highest SFs sampled are steadily increasing across time (a2 = 3.68 cpd; b2 = 10.32 cpd/s) and in
which, perhaps more surprisingly, the lowest SFs sampled are the same throughout the video (a1
= 0.69 cpd; b1 = 0.00 cpd/s; Figure 2).

Hand Position & Spatial Frequency Use
9
In summary, the observed time course of SF sampling matches a coarse-to-fine model,
thus confirming what has been observed in previous studies. Further, our method characterized
this sampling pattern with greater precision than previous methods and showed that low SFs
continue to be used in the latest time frames (see also Caplette et al., 2016). Together, these
results demonstrate the value and the reliability of our method.
Experiment 2
In Experiment 2, we employed the technique that was validated in Experiment 1 to
investigate with unprecedented precision how hand position (i.e. when hands are near or far from
the stimulus) modulates the time course of SF sampling. We expected to replicate, with everyday
objects, the finding that the proximity of the hands to the stimulus enhances the extraction of low
SFs and/or impair the extraction of high SFs reported in the literature . Furthermore, we believed
that the high SF resolution of our method would allow us to detect the precise SFs affected by
hand position, and that its high temporal resolution would allow us to discover the precise
moments during object recognition at which hand position influences SF processing.
Method
Twenty-eight right-handed adult participants (11 males; mean age = 22.1, SD = 2.19)
were recruited on the campus of the University of Montreal. Subjects had normal or corrected to
normal vision, and did not suffer from any visual or reading disability. The study was approved
by the ethics board of the University of Montreal's Faculty of Arts and Sciences. Written consent
from all participants was obtained after the procedure had been fully explained, and a monetary
compensation was provided upon completion of the experiment.
During the task, chin rests were used to maintain viewing distance at 35 cm; images
subtended 13 x 13 degrees of visual angle. Importantly, half the trials were performed with a

Hand Position & Spatial Frequency Use
10
keyboard key press (hands-distal condition), and half were performed with two mice attached to
either side of the monitor (hands-proximal condition; see Gozli et al., 2012). Participants' elbows
were resting on the table while in the hands-proximal condition so that no physical effort had to
be exerted. Conditions were alternated in blocks of 50 trials (the first condition was
counterbalanced among participants).
Results and discussion
Participants responded correctly on an average of 73.13% of the trials in the hands-
proximal condition, and of 74.10% in the hands-distal condition (t(27) = 0.96, p > .25); they
required an average of 66.26 bubbles in the hands-proximal condition, and of 66.93 bubbles in
the hands-distal condition (t(27) = 0.84, p > .25). In agreement with a previous study, the mean
response time was shorter in the hands-proximal condition compared to the hands-distal
condition (633 ms vs 747 ms; t(27) = 4.26, p < .001).
Figure 3 illustrates the z-scored group classification images for the two conditions and the
contrast between them. Z-scores indicate the correlation between the presentation of a given SF
on a given time frame and accuracy; white curves indicate significant clusters (p < .05, two-
tailed, FWER-corrected). In the hands-proximal condition, a first cluster that peaked at 1.27 cpd
and 88 ms (Zmax = 5.08, k = 391) and a second cluster that peaked at 0.35 cpd and 296 ms (Zmax =
4.17, k = 78) led to accurate object recognition, while a third cluster that peaked at 4.31 cpd and
113 ms (Zmax = 4.60, k = 81) led to inaccurate object recognition. In the hands-distal condition, a
unique cluster that peaked at 1.15 cpd and 46 ms (Zmax = 4.63, k = 191) led to statistically
significant accurate object recognition. This is very similar to the early SF sampling pattern
observed in Experiment 1.
In the contrast between these two conditions, one cluster that peaked at 0.27 cpd and 288
ms (Zmax = 4.26, k = 51) led to more accurate recognition in the hands-proximal condition than in

Hand Position & Spatial Frequency Use
11
the hands-distal condition, while a second cluster that peaked at 4.42 cpd and 104 ms (Zmax =
4.46, k = 124) led to more accurate recognition in the hands-distal condition relative to the hands-
proximal condition.
Given the fact that each object was repeated on average 23 times during the course of the
experiment (although always with different SFs revealed at different moments), we tested
whether there was some learning effect. To do that, we contrasted classification images derived
from the first and last blocks of trials. We did not find any significant difference; note however
that this result should be interpreted carefully given the poor signal-to-noise ratio in our data.
In summary, we showed that the sampling of relatively high SFs peaking at 4.42 cpd is
impaired and that the sampling of relatively low SFs peaking at 0.27 cpd is enhanced when
objects are near the hands. Most importantly, by evaluating the time course of SF sampling when
hands were near target objects, we showed that the bias towards low SF processing occurs in the
latest time frames at around 288 ms, while the decreased sensitivity to high SFs occurs around
104 ms.
General Discussion
The main goal of the present study was to investigate how the time course of SF sampling
is altered when objects are presented near the hands. On each trial, subjects had to recognize an
object from a brief video sampling random SFs on random frames; we then reverse correlated the
revealed SFs and time frames with response accuracy. This technique allowed us to map the time
course of SF sampling with unprecedented precision.
We first put our method to the test by examining the time course of SF sampling in a basic
object recognition task. As expected, we observed the classic coarse-to-fine sampling reported in
the literature (Caplette et al., 2016; Hughes et al., 1996; Hupe et al., 2001; Schyns & Oliva,

Hand Position & Spatial Frequency Use
12
1994). However, our method allowed us to characterize this coarse-to-fine sampling with much
greater precision than previous methods, notably indicating that low SFs are used continuously.
These results demonstrate the value and the reliability of our method.
In our second experiment, we tackled our main research question: How exactly does hand
position alter SF sampling? We replicated the finding — and extended it to everyday objects —
that the prioritization of objects near the hands is driven by an increased use of relatively low SFs
and a decreased use of relatively high SFs when hands were proximal to the target object.
Importantly, our high-resolution technique provided the increased resolution to reveal that this
effect is driven specifically by low SFs peaking at 0.27 cpd and high SFs peaking at 4.42 cpd.
These results are consistent with a biasing of processing toward magnocellular pathways when
hands are near the stimuli.
Most importantly, this technique gave us a novel opportunity to examine the time course
of SF use as a function of hand position. In both hands-proximal and -distal conditions, low SFs
in early stages of object recognition (peaking at 46 ms and 88 ms) contributed to accurate object
recognition, while high SFs presented around 113 ms led to decreased accuracy in the hands-
proximal condition. In later stages of object recognition (around 288 ms), low SFs contributed
significantly more to accurate object identification in the hands-proximal condition than in the
hands-distal condition.
The time course of the effect of hands on SF sampling informs us about underlying object
recognition mechanisms. The discovery that hand position modulates SF sampling in later time
frames (> 100 ms) suggests that the effect is attentional rather than purely perceptual. The fact
that hand position modulates the use of high SFs seen around 113 ms and low SFs seen around
288 ms in the videos implies that this information is processed by the brain later than these
latencies. This is relatively late by object recognition standards: the first bottom-

Hand Position & Spatial Frequency Use
13
up object processing sweep is believed to extend up to about 100 ms after stimulus onset (Lamme
& Roelfsema, 2000). Thus, the effect of hand position on SF processing appears to have a top-
down component, which involves attentional selection of visual information. Some researchers
have already proposed that space near the hands is attentionally prioritized (Abrams et al., 2008;
Reed et al., 2006, 2010); our findings furthermore elucidate that specific SFs are prioritized or
inhibited in the near-hands space. This finding reconciles attentional and magnocellular accounts
of the hands effect: attention acts on specific SFs by biasing processing toward the magnocellular
or parvocellular pathway (attention can exert its influence as early as the LGN; e.g., O’Connor et
al., 2002; McAlonan et al., 2008). Faster processing in near-hands space (e.g., Reed et al., 2006)
might be due to this biasing toward the magnocellular pathway, which conducts information at a
faster rate (see Gozli et al., 2012).
Further, the recently discovered interaction between the attentional demands of a given
task and the SFs modulated by hand position also supports the hypothesis that the effect of hands
on SF use is attentional (Goodhew & Clarke, 2016). Future studies using this new dynamic
stimulus presentation method could help confirm this conclusion. For example, both the
attentional demands and hand position could be manipulated (as in Goodhew & Clarke, 2016)
and the similarity of the time frames of the effects of both factors could be assessed. Relatedly,
we could also evaluate the time course of SF use in a condition that emphasizes top-down
processing and in another that emphasizes bottom-up processing (e.g., through priming or not the
object identity before the stimulus). By verifying if the hand position effect can be explained by
the effect of either attentional condition, we could disentangle these two explanations;
furthermore, this would provide a powerful test of popular object recognition models (e.g., Bar,
2003; Bullier, 2001).

Hand Position & Spatial Frequency Use
14
In conclusion, our results demonstrate that the visual system biases processing in
magnocellular and parvocellular pathways according to hand position at a late processing stage.
Using the method introduced in this paper, future studies can examine how the hand-position
phenomenon interacts with different attentional demands.

Hand Position & Spatial Frequency Use
15
References
Abrams, R. A., Davoli, C. C., Du, F., Knapp, W. H., 3rd, & Paull, D. (2008). Altered vision near
the hands. Cognition, 107(3), 1035-1047. doi: 10.1016/j.cognition.2007.09.006
Abrams, R. A., & Weidler, B. J. (2014). Trade-offs in visual processing for stimuli near the
hands. Atten Percept Psychophys, 76(2), 383-390. doi: 10.3758/s13414-013-0583-1
Allard, R., & Faubert, J. (2008). The noisy-bit method for digital displays: Converting a 256
luminance resolution into a continuous resolution. Behavior Research Methods, 40(3), 735–
743. http://doi.org/10.3758/BRM.40.3.735
Bar, M., Kassam, K. S., Ghuman, A. S., Boshyan, J., Schmid, A. M., Dale, A. M., . . . Halgren, E.
(2006). Top-down facilitation of visual recognition. Proc Natl Acad Sci U S A, 103(2), 449-
454. doi: 10.1073/pnas.0507062103
Bar, M. (2003). A Cortical Mechanism for Triggering Top-Down Facilitation in Visual Object
Recognition. Journal of Cognitive Neuroscience, 15(4), 600–609.
http://doi.org/10.1126/science.8316836
Bekkering, H., & Neggers, S. F. (2002). Visual search is modulated by action intentions. Psychol
Sci, 13(4), 370-374.
Brainard, D. H. (1997). The Psychophysics Toolbox. Spatial Vision, 10(4), 433–436.
Brown, L. E., Morrissey, B. F., & Goodale, M. A. (2009). Vision in the palm of your hand.
Neuropsychologia, 47(6), 1621-1626. doi: 10.1016/j.neuropsychologia.2008.11.021
Bullier, J. (2001). Integrated model of visual processing. Brain Res Brain Res Rev, 36(2-3), 96-
107.
Callaway, E. M. (1998). Local circuits in primary visual cortex of the macaque monkey. Annu
Rev Neurosci, 21, 47-74. doi: 10.1146/annurev.neuro.21.1.47
Caplette, L., West, G., Gomot, M., Gosselin, F., & Wicker, B. (2014). Affective and contextual
values modulate spatial frequency use in object recognition, 5:512.
http://doi.org/10.3389/fpsyg.2014.00512/abstract
Caplette, L., Wicker, B., & Gosselin, F. (2016). Atypical Time Course of Object Recognition in
Autism Spectrum Disorder. Scientific reports, 6.
Chauvin, A., Worsley, K. J., Schyns, P. G., Arguin, M., & Gosselin, F. (2005). Accurate
statistical tests for smooth classification images. Journal of Vision, 5(9), 659–667.
Davoli, C. C., Du, F., Montana, J., Garverick, S., & Abrams, R. A. (2010). When meaning
matters, look but don't touch: the effects of posture on reading. Mem Cognit, 38(5), 555-
562. doi: 10.3758/MC.38.5.555

Hand Position & Spatial Frequency Use
16
De Valois, R. L., Albrecht, D. G., & Thorell, L. G. (1982). Spatial frequency selectivity of cells
in macaque visual cortex. Vision Res, 22(5), 545-559.
Derrington, A. M., & Lennie, P. (1984). Spatial and temporal contrast sensitivities of neurones in
lateral geniculate nucleus of macaque. J Physiol, 357, 219-240.
Fagioli, S., Hommel, B., & Schubotz, R. I. (2007). Intentional control of attention: action
planning primes action-related stimulus dimensions. Psychol Res, 71(1), 22-29. doi:
10.1007/s00426-005-0033-3
Goodale, M. A., & Milner, A. D. (1992). Separate visual pathways for perception and action.
Trends Neurosci, 15(1), 20-25.
Goddard, E., Carlson, T. A., Dermody, N., & Woolgar, A. (2016). Representational dynamics of
object recognition: Feedforward and feedback information flows. NeuroImage, 128, 385–
397. http://doi.org/10.1016/j.neuroimage.2016.01.006
Goodhew, S. C., & Clarke, R. (2016). Contributions of parvocellular and magnocellular
pathways to visual perception near the hands are not fixed, but can be dynamically
altered. Psychon Bull Rev, 23(1), 156-162. doi: 10.3758/s13423-015-0844-1
Goodhew, S. C., Edwards, M., Ferber, S., & Pratt, J. (2015). Altered visual perception near the
hands: A critical review of attentional and neurophysiological models. Neurosci Biobehav
Rev, 55, 223-233. doi: 10.1016/j.neubiorev.2015.05.006
Goodhew, S. C., Fogel, N., & Pratt, J. (2014). The nature of altered vision near the hands:
evidence for the magnocellular enhancement account from object correspondence through
occlusion. Psychon Bull Rev, 21(6), 1452-1458. doi: 10.3758/s13423-014-0622-5
Gozli, D. G., West, G. L., & Pratt, J. (2012). Hand position alters vision by biasing processing
through different visual pathways. Cognition, 124(2), 244-250. doi:
10.1016/j.cognition.2012.04.008
Hughes, H. C., Nozawa, G., & Kitterle, F. (1996). Global precedence, spatial frequency channels,
and the statistics of natural images. J Cogn Neurosci, 8(3), 197-230. doi:
10.1162/jocn.1996.8.3.197
Hupe, J. M., James, A. C., Girard, P., Lomber, S. G., Payne, B. R., & Bullier, J. (2001). Feedback
connections act on the early part of the responses in monkey visual cortex. J Neurophysiol,
85(1), 134-145.
Jemel, B., Mimeault, D., Saint-Amour, D., Hosein, A., & Mottron, L. (2010). VEP contrast
sensitivity responses reveal reduced functional segregation of mid and high filters of visual
channels in autism. Journal of Vision, 10(6):13, 1–13. http://doi.org/10.1167/10.6.13
Kaplan, E., & Shapley, R. M. (1986). The primate retina contains two types of ganglion cells,
with high and low contrast sensitivity. Proc Natl Acad Sci U S A, 83(8), 2755-2757.

Hand Position & Spatial Frequency Use
17
Keita, L., Guy, J., Berthiaume, C., Mottron, L., & Bertone, A. (2014). An early origin for detailed
perception in Autism Spectrum Disorder: biased sensitivity for high-spatial frequency
information. Sci Rep, 4, 5475. doi: 10.1038/srep05475
Lamme, V. A. F., & Roelfsema, P. R. (2000). The distinct modes of vision offered by
feedforward and recurrent processing. Trends in Neuroscience, 23(11), 571–579.
Livingstone, M., & Hubel, D. (1988). Segregation of form, color, movement, and depth:
anatomy, physiology, and perception. Science, 240(4853), 740-749.
McAlonan, K., Cavanaugh, J., & Wurtz, R. H. (2008). Guarding the gateway to cortex with
attention in visual thalamus. Nature, 456(7220), 391–394.
http://doi.org/10.1038/nature07382
O'Connor, D. H., Fukui, M. M., Pinsk, M. A., & Kastner, S. (2002). Attention modulates
responses in the human lateral geniculate nucleus. Nature Neuroscience, 5(11), 1203–1209.
http://doi.org/10.1038/nn957
Parker, D. M., & Salzen, E. A. (1977). The spatial selectivity of early and late waves within
the human visual evoked response. Perception, 6:85-95.
Pelli, D. G. (1997). The VideoToolbox software for visual psychophysics: transforming numbers
into movies. Spatial Vision, 10(4), 437–442.
Reed, C. L., Grubb, J. D., & Steele, C. (2006). Hands up: attentional prioritization of space near
the hand. J Exp Psychol Hum Percept Perform, 32(1), 166-177. doi: 10.1037/0096-
1523.32.1.166
Reed, C. L., Betz, R., Garza, J. P., & Roberts, R. J. (2010). Grab it! Biased attention in functional
hand and tool space. Attention, Perception, & Psychophysics, 72(1), 236–245.
http://doi.org/10.3758/APP.72.1.236
Schyns, P. G., & Oliva, A. (1994). From blobs to boundary edges: Evidence for time-and spatial-
scale-dependent scene recognition. Psychol Sci, 5, 195-200.
Shapley, R. (1990). Visual sensitivity and parallel retinocortical channels. Annu Rev Psychol, 41,
635-658. doi: 10.1146/annurev.ps.41.020190.003223
Shenhav, A., Barrett, L. F., & Bar, M. (2012). Affective value and associative processing share a
cortical substrate. Cognitive, Affective, and Behavioral Neuroscience, 13(1), 46–59.
http://doi.org/10.3758/s13415-012-0128-4
Willenbockel, V., Sadr, J., Fiset, D., Horne, G. O., Gosselin, F., & Tanaka, J. W. (2010).
Controlling low-level image properties: The SHINE toolbox. Behavior Research Methods,
42(3), 671–684. http://doi.org/10.3758/BRM.42.3.671

Hand Position & Spatial Frequency Use
18
Wohlschlager, A. (2000). Visual motion priming by invisible actions. Vision Res, 40(8), 925-
930.

Hand Position & Spatial Frequency Use
19
Figure Captions
Figur e 1. Illustration of the sampling method. On each trial, we randomly generated a matrix of
dimensions 256 x 40 (representing respectively SFs and frames) in which most elements were
zeros and a few were ones. We then convolved this sparse matrix with a 2D Gaussian kernel (a
"bubble"). This resulted in the trial's sampling matrix, shown here as a plane with a number of
randomly located bubbles. Every column of this sampling matrix was then rotated around its
origin to create isotropic 2D random filters. Finally, these 2D random filters were dot-multiplied
by the base image's spectrum and inverse fast Fourier transformed to create a filtered version of
the image for every video frame.
Figure 2. Classification image depicting the correlations between SF-time pixels and accurate
object recognition. Pixels enclosed by solid lines are significant (p < .05, two-tailed, FWER-
corrected). Dashed lines represent the best fitting linear SF sampling model (see text for details).
Figure 3. Group classification images depicting the correlations between SF-time pixels and
accurate object recognition: a) Hands-Proximal condition; b) Hands-Distal condition; c) Hands-
Proximal condition – Hands-Distal condition. Pixel clusters enclosed by white lines are
significant (p < .05, two-tailed, FWER-corrected).

Hand Position & Spatial Frequency Use
20
Figure 1.

Hand Position & Spatial Frequency Use
21
Figure 2

Hand Position & Spatial Frequency Use
22
Figure 3.
Time (ms)
50 100 150 200 250 300
1
2
3
4
5
6
7
8
9
Time (ms)
50 100 150 200 250 300
Spatial Frequency (cpd)
1
2
3
4
5
6
7
8
9
Time (ms)
50 100 150 200 250 300
1
2
3
4
5
6
7
8
9
-4
-3
-2
-1
0
1
2
3
4
5
Z
abc