Download Is In-out asymmetry diagnostic of visual crowding? Ramakrishna

Is In-out asymmetry diagnostic of visual crowding?
Ramakrishna Chakravarthi and Alasdair D F Clarke
Objects in clutter are hard to identify. This clutter induced interference is known as visual
crowding. Visual crowding has been extensively studied over the past several decades and has
provided valuable insights into several cognitive processes, including object recognition (Levi,
2008; Whitney & Levi, 2011), reading (Martelli, Di Filippo, Spinelli, & Zoccolotti, 2009; Pelli et al.,
2007), and awareness (Atas, Faivre, Timmermans, Cleeremans, & Kouider, 2014; Kouider,
Berthet, & Faivre, 2011). However, deleterious interactions between an object and its flankers
can also occur due to other spatial processes, including surround suppression and contrast
masking (Levi, Hariharan, & Klein, 2002; Pelli, Palomares, & Majaj, 2004; Petrov, Popple, &
McKee, 2007), wherein nearby flankers suppress the target. It has been suggested (Levi et al.,
2002; Pelli et al., 2004) that crowding and contrast masking mechanisms can be distinguished by
a set of properties: masking impairs detection, whereas crowding only affects identification;
masking scales with the size of the objects, whereas crowding does not; masking is independent
of eccentricity, whereas crowding is deeply dependent on it. Hence, the diagnostic test to
determine if a particular target has suffered from crowding and not from masking was to check
if the spatial extent of the target-flanker interaction scaled with eccentricity but not with
stimulus size. If it did, then it was crowding. However, this set of diagnostic properties was
challenged by Petrov et al. (2007), who argued that surround suppression also has the same set
of features. They proposed that the unique property differentiating crowding and surround
suppression was that crowding displayed a specific sort of asymmetry known as In-Out
Asymmetry (IOA) that surround suppression did not. IOA is the finding that the flanker farther
away from the fovea (the ‘outward’ flanker) relative to the target causes more crowding than
the flanker that is closer to the fovea (the ‘inward’ flanker). This appears counter-intuitive since
the inward flanker should be more ‘perceptible’, because of better acuity for objects closer to
the fovea, and hence should lead to more interference. However, increased interference from
the outward flanker can be explained by cortical magnification, due to which neurons that fire
to the outward flanker are closer to the neurons responding to the target than are the neurons
responding to the inward flanker (Pelli, 2008), and hence can cause more interference. IOA has
been noticed and reported in a variety of settings right from the early days of crowding research
(Bex, Dakin, & Simmers, 2003; Bouma, 1970; Manassi, Sayim, & Herzog, 2012). Consequently,
Petrov et al. (2007) proposed that demonstration of IOA should be considered diagnostic of
crowding.
However, IOA is not observed in all situations. For example, Petrov and Meleshkevich
(2011a) found that, in the specific setup they used, IOA was present only in the horizontal
meridian and not at other locations. In fact, most previous studies where IOA has been
demonstrated also seem to have documented it along the horizontal axis; few have tested other
locations. They also showed that IOA was stronger if participants had to monitor two spatial
locations than if the target was consistently presented at the same location. They argued that
this effect could be ascribed to attentional deployment. That is, IOA is observed only if attention
is spread out, but not if it can be focused on a given location, which is the case when the target
is predictably presented at the same location. Further evidence for the role of attention was
provided when, in a different study, they showed that varying the target location, and thus
increasing the need for attention to monitor more locations, increased the strength of IOA.
Additionally, when participants had to attend a central location, IOA was eliminated or even
reversed in some participants (Petrov & Meleshkevich, 2011b). On the other hand, there are
potential issues with the stimulus setup and participants used in these paradigms. Generally,
these studies have been conducted with a few, experienced participants. Further, considerable
inter-observer variability has been documented. A property should be evident consistently in all
observers and not just experienced ones if it is to be considered diagnostic. Incidentally, Petrov
and Meleshkevich (2011b, 2011a) used coarsely discriminable stimuli (identifying Gabor
orientations separated by 90 deg), which themselves might result in little to no crowding (Pelli
et al., 2004). Hence, stimulus choice might also be important in determining the robustness of
IOA. Given these findings, the question arises if IOA exists and if it can be considered diagnostic
of crowding.
This study, therefore, is intended to address a few related issues: 1) Does IOA exist? 2) If
yes, can it be observed in a relatively larger sample of naïve observers and how prevalent is it
within this sample? 3) The strength of IOA appeared to vary dramatically. Here we hope to
derive an estimate of the strength of IOA and its variance across the sample tested. 4) Is IOA
restricted to the horizontal meridian, as claimed by Petrov and Meleshkevich (2011a)? If so, it
cannot be considered diagnostic of crowding. 5) It was suggested that IOA depended on
attentional deployment. This study will test if this is the case by directly manipulating
attentional deployment using precues in a typical crowding paradigm. Finally, we will use letters
as stimuli, which require fine discrimination and have been widely documented to be
susceptible to crowding.
Methods:
40 undergraduate students will be recruited at two separate sites (20 each at Universities
of Aberdeen and Essex). Stimuli will be generated using MATLAB with Psychtoolbox extensions
(Kleiner, Brainard, & Pelli, 2007) and presented on CRT monitors in a darkened room. Stimuli will
be 9 black letters in the Sloan font (D, H, K, N, O, R, S, V and Z) presented on a grey background.
The letter C is omitted since C and O are much less discriminable than other pairs in the Sloan
font (Elliott, Whitaker, & Bonette, 1990). One target and one flanker from this set will be
randomly chosen on any given trial. The target will be presented at 8 deg eccentricity at one of
four locations (left, right, above, or below the fixation, on the horizontal or vertical meridian).
Two white lines, 2 deg on either side of the meridian, will mark the position of each target
location throughout the experiment, so as to reduce confusion about which of the two letters is
the target. These lines should not cause any crowding themselves, as they will be placed far
apart in the tangential direction and will be of the opposite contrast polarity to the target (Kooi,
Toet, Tripathy, & Levi, 1994; Toet & Levi, 1992).
Participants will be tested in three conditions: Blocked target location, Random target
location, and Random target location with precue. In the blocked target location condition, the
target will always be presented at the same location pre-chosen from the four possible
locations. In the random target location condition, the target can be presented in any of the
four locations with equal probability. Finally, the random target location with precue condition
will utilise the same procedure as the previous condition, but with a precue (a 0.3 deg red dot)
presented at the target location 100 ms before target onset (Nakayama & Mackeben, 1989).
The cue will last for 50 ms. The target might be accompanied by a single flanker, placed either
inwards or outwards relative to the target. The target (and flanker) will be presented for 100
ms. The participant will then be asked to report the identity of the target letter. No feedback
will be provided. Overall, we will assess performance in three conditions, four target locations
and two flanker locations, for a total of 24 combinations.
Figure 1: Stimulus conditions in experiment 1. Panel (a) illustrates trials in the ‘Blocked target
location’ condition. Here, targets are always presented at the same location within a block. In this
example, a single ‘outward’ flanker is presented along with the target. However, in other blocks, the
flanker might be presented on the ‘inward’ side or not presented. The size of the letters (and hence
spacing, which is set as 1.1*size) is controlled by the QUEST algorithm to determine flanked acuity.
Panel (b) illustrates trials in the ‘Random target location’ condition, where targets and their flankers
can be presented in any of the four locations within the block. Finally, panel (c) illustrates trials in the
‘Random target location with precue’ condition where a target is presented at a randomly chosen but
precued location. Panels a and b show outward flankers, and panel c shows inward flankers.
Following Song, Levi and Pelli (2014) and Petrov and Meleshkevich (2011b, 2011a), we will
compare flanked with unflanked visual acuity to determine the extent of crowding for each
combination. We will obtain, separately, flanked and unflanked acuity thresholds using the
QUEST algorithm. This algorithm will control the size of an isolated letter to estimate unflanked
acuity and it will control the sizes of both the target and its flanker to obtain flanked acuity. In
the latter case, the centre-to-centre spacing between the target and the flanker will be fixed at
1.1 times the size, such that the spacing increases with size, in order to prevent overlap and to
measure the extent of crowding. Each acuity (flanked and unflanked for each condition-target
location-flanker location combination) will be assessed twice with a run of 40 trials each and
averaged. The ratio of flanked to unflanked acuity for each combination, called the crowding
factor (Petrov & Meleshkevich, 2011b, 2011a), will specify the extent of crowding. A crowding
factor of greater than one will indicate the presence of crowding, and the larger the factor the
larger the target-flanker interference zone is. We will then take a ratio of outward to inward
crowding factors. A ratio of greater than one, here, will indicate IOA. This ratio can be used to
determine the answers to the five questions posed above.
In sum, we will determine 48 flanked acuity thresholds and 24 unflanked thresholds.
These will be randomly intermixed and distributed over two sessions of about 45 min each per
participant.
The parameters proposed here (target eccentricity, target location marker size and
configuration, stimulus duration, cue duration, colour and size, etc.) are based on reasonable
estimates derived from previous successful crowding experiments. However, these might not be
optimal for testing our questions. We will initially run 4-5 pilot participants with these
parameters and ascertain if we can obtain meaningful thresholds. If we do not, then we will
modify the parameters accordingly. Nevertheless, we plan to also report what didn’t work.
Experiment 2: If we discover an effect of attentional deployment in experiment 1, we will
run a second experiment that will more systematically vary deployment of both endogenous
and exogenous attention to test if either kind of attention influences IOA. Here, we will use 8
possible target locations arranged equidistantly along an imaginary circle around fixation (radius
8 deg), indicated by pairs of lines (their orientations are changed relative to experiment 1,
because using the same orientations as in that experiment would cause confusion due to the
extra locations coupled with grouping due to proximity of lines belonging to adjacent targets).
We will then cue 1, 2, 4 or all 8 locations, either endogenously (white arrows pointing to the
locations with an onset 350 ms before target onset) or exogenously (a red dot at the relevant
target locations 100 ms before target onset). Subsequently a target and one flanker (inward or
outward) will be presented at one of the cued locations. We will determine crowding factors
under these conditions and check if IOA is affected by the number of cued locations and the
type of cueing.
Figure 2: Protocol for Experiment 2. In this experiment, there will be eight target locations,
indicated by pairs of white lines. 1, 2, 4 or 8 locations will be precued either endogenously (panel a) or
exogenously (panel b). The target with its flanker (inward, outward or none) will be presented in one
of the cued locations.
Acknowledgements
We thank John Greenwood and Mauro Manassi for valuable comments on a previous
version of this proposal.
References
Atas, A., Faivre, N., Timmermans, B., Cleeremans, A., & Kouider, S. (2014). Nonconscious
learning from crowded sequences. Psychological Science, 25(1), 113–119.
Bex, P. J., Dakin, S. C., & Simmers, A. J. (2003). The shape and size of crowding for moving
targets. Vision Research, 43(27), 2895–2904. https://doi.org/10.1016/S00426989(03)00460-7
Bouma, H. (1970). Interaction Effects in Parafoveal Letter Recognition. Nature, 226, 177–178.
https://doi.org/10.1038/226177a0
Elliott, D. B., Whitaker, D., & Bonette, L. (1990). Differences in the legibility of letters at contrast
threshold using the Pelli-Robson chart. Ophthalmic and Physiological Optics, 10(4), 323–
326. https://doi.org/10.1111/j.1475-1313.1990.tb00877.x
Kleiner, M., Brainard, D., & Pelli, D. G. (2007). What’s new in Psychtoolbox-3? Perception,
36(14), 1.
Kooi, F. L., Toet, A., Tripathy, S. P., & Levi, D. M. (1994). The effect of similarity and duration on
spatial interaction in peripheral vision. Spatial Vision, 8(2), 255–279.
Kouider, S., Berthet, V., & Faivre, N. (2011). Preference Is Biased by Crowded Facial Expressions.
Psychological Science, 22(2), 184–189. https://doi.org/10.1177/0956797610396226
Levi, D. M. (2008). Crowding—An essential bottleneck for object recognition: A mini-review.
Vision Research, 48(5), 635–654. https://doi.org/10.1016/j.visres.2007.12.009
Levi, D. M., Hariharan, S., & Klein, S. A. (2002). Suppressive and facilitatory spatial interactions in
peripheral vision: Peripheral crowding is neither size invariant nor simple contrast
masking. Journal of Vision, 2(2), 3–3.
Manassi, M., Sayim, B., & Herzog, M. H. (2012). Grouping, pooling, and when bigger is better in
visual crowding. Journal of Vision, 12(10), 13–13. https://doi.org/10.1167/12.10.13
Martelli, M., Di Filippo, G., Spinelli, D., & Zoccolotti, P. (2009). Crowding, reading, and
developmental dyslexia. Journal of Vision, 9(4), 14–14. https://doi.org/10.1167/9.4.14
Nakayama, K., & Mackeben, M. (1989). Sustained and transient components of focal visual
attention. Vision Research, 29(11), 1631–1647. https://doi.org/10.1016/00426989(89)90144-2
Pelli, D. G. (2008). Crowding: a cortical constraint on object recognition. Current Opinion in
Neurobiology, 18(4), 445–451. https://doi.org/10.1016/j.conb.2008.09.008
Pelli, D. G., Palomares, M., & Majaj, N. J. (2004). Crowding is unlike ordinary masking:
Distinguishing feature integration from detection. Journal of Vision, 4(12), 12.
https://doi.org/10.1167/4.12.12
Pelli, D. G., Tillman, K. A., Freeman, J., Su, M., Berger, T. D., & Majaj, N. J. (2007). Crowding and
eccentricity determine reading rate. Journal of Vision, 7(2), 20.
https://doi.org/10.1167/7.2.20
Petrov, Y., & Meleshkevich, O. (2011a). Asymmetries and idiosyncratic hot spots in crowding.
Vision Research, 51(10), 1117–1123. https://doi.org/10.1016/j.visres.2011.03.001
Petrov, Y., & Meleshkevich, O. (2011b). Locus of spatial attention determines inward-outward
anisotropy in crowding. Journal of Vision, 11(4), 1–1. https://doi.org/10.1167/11.4.1
Petrov, Y., Popple, A. V., & McKee, S. P. (2007). Crowding and surround suppression: Not to be
confused. Journal of Vision, 7(2), 12. https://doi.org/10.1167/7.2.12
Song, S., Levi, D. M., & Pelli, D. G. (2014). A double dissociation of the acuity and crowding limits
to letter identification, and the promise of improved visual screening. Journal of Vision,
14(5), 3–3. https://doi.org/10.1167/14.5.3
Toet, A., & Levi, D. M. (1992). The Two-Dimensional Shape of Spatial. Retrieved from
http://www.academia.edu/download/41763762/toet_levi1992.pdf
Whitney, D., & Levi, D. M. (2011). Visual crowding: a fundamental limit on conscious perception
and object recognition. Trends in Cognitive Sciences, 15(4), 160–168.
https://doi.org/10.1016/j.tics.2011.02.005