Extensive psychophysical studies have established that adult spatial vision is fundamentally limited by crowding—the reduced ability to identify an object in the periphery when it is surrounded by other objects (Whitney and Levi, 2011
). This limit extends beyond that of visual acuity, because even though the object is still visible, it is not readily identifiable.
Crowding is the operational measure of spatial resolution of attention. When multiple objects fall within a peripheral region of the visual field, features within and between the objects appear jumbled, impairing identification of the target object (He et al., 1996
; Cavanagh et al., 1999
; Intriligator and Cavanagh, 2001
). Crowding has been demonstrated using gratings, numbers, letters and faces. In the periphery, spatial resolution decreases with increasing eccentricity and with reduced target-flanker spacing. Crowding is thought to be minimal or non-existent at the fovea (Bouma, 1970
; Bouma and Andriessen, 1970
; Andriessen and Bouma, 1976
; Tripathy and Cavanagh, 2002
; Pelli et al., 2004
; Louie et al., 2007
; Levi, 2008
; Pelli and Tillman, 2008
Here, we utilized our recently developed eye-tracking paradigm (Farzin et al., 2010
) to measure the extent of spatial crowding in infants with fragile X syndrome compared with developmental age-matched neurotypical infants.
Participants included in this experiment were 32 infants diagnosed with the fragile X syndrome full mutation (13 females, mean chronological age = 26.10 ± 13.15 months, range = 6–46 months). For this and the following experiments, infants with fragile X syndrome were recruited from and clinically evaluated at the MIND Institute Fragile X Research and Treatment Centre as part of a larger longitudinal study on visual and cognitive development of infants with fragile X syndrome. Infants with the disorder underwent a routine blood test to confirm the presence of the full mutation (>200 CGG repeats). Genomic DNA was isolated from peripheral blood lymphocytes using standard methods (Puregene Kit; Gentra Inc.), and analysis and calculation of the CGG repeat size were carried out using an Alpha Innotech FluorChem 8800 Image Detection System (Tassone et al., 2000
messenger RNA level was quantified using a 7900 Sequence detector (PE Biosystems) as previously described (Tassone et al., 2000
). Since larger expansions of CGG repeats (>200) typically result in reduced levels or absence of FMR1
messenger RNA and the gene's protein product, FMRP, we examined the relationship between these molecular measures and performance on the psychophysical tasks to gain further insight into the role of FMRP in brain development.
The developmental level of infants with fragile X syndrome was evaluated by a trained researcher using the Mullen Scales of Early Learning standardized assessment (MSEL; Mullen, 1995
), and revealed a mean developmental age of 17.23 months [standard deviation (SD) = ± 10.24 months, range = 4–32 months]. Infants with fragile X syndrome were matched based on their developmental age to a group of 32 neurotypical infants, who were chronologically younger in age than the participants with fragile X syndrome (10 females, mean chronological age = 15.28 ± 5.69 months, range = 6–22 months). There was no significant difference in developmental level between the neurotypical controls and the infants with fragile X syndrome, based on the assumption that neurotypical infants’ developmental age was approximately equivalent to their chronological age [F
(1, 63) = 2.78, P =
0.101]. Neurotypical infants were recruited through letters to families, fliers and word of mouth in Davis, California and surrounding areas. All participants had uncorrected visual acuity. The study was approved by the Institutional Review Board at the University of California, Davis, and informed consent was obtained from a parent or legal guardian of all infants.
Apparatus and stimuli
The experiment has been described in detail previously (Farzin et al., 2010
). The experiment was programmed in Presentation version 11.3 (Neurobehavioral Systems, Inc.) and stimuli were presented on a Tobii 17-inch LCD binocular eye tracker monitor.
Faces were 99.77% Michelson contrast, and were cropped to fit into a 5° by 3° ellipse when viewed from a distance of 60 cm. Six 1.53° by 1.05° flankers were created by ‘cutting’ elliptically shaped sections from each upright target face. In the crowded condition, flankers were presented surrounding the target faces at a fixed horizontal centre-to-centre distance of 2.2° between the target face and the flanker. Stimuli were presented against a grey background (77.23 cd/m2; ).
(A) Mooney face stimuli used in Experiment 1; (B) upright face without flankers; (C) inverted face without flankers; (D) upright face with flankers; and (E) inverted face with flankers.
All experiments began with a five-point calibration routine on the eye tracker to estimate the infant's gaze position accurately during the task. Trials began with a 1° central fixation video until the infant's gaze was obtained within a 1° radius around the video. Trials in which an infants’ central fixation was not obtained within 10 s, or which shifted outside of the radius before the faces were presented [on average four trials per infant, no group difference [t(62) = 1.52, P = 0.134], were discarded. Immediately following the central fixation video, one upright and one inverted Mooney face (the same identity) were shown, one face to the left and one to the right of fixation at a centre-to-centre distance of either 3, 6 or 10° along the horizontal meridian. Both faces were presented with (crowded) or without (uncrowded) the six corresponding flanker parts for 2 s (). The eccentricity at which the faces were presented was blocked in sets of four trials, and the order of blocks, side of upright face and the presence of flankers were randomized.
Data coding and threshold estimation
Eye-tracking data were coded offline using Noldus Observer 5.0 software, and statistical analyses were performed with SPSS (version 16.0). The primary measure of performance on each trial was the location of infants’ first fixation from the central video immediately following the onset of the faces. A fixation was defined as a series of data points within a 30 pixel radius for a minimum duration of 100 ms. A first fixation was coded as a hit (1) if it was on the upright face and a miss (0) if it was on the inverted face. In order to make a first saccade that landed on the upright face, the infant must have perceived the upright face in the periphery. If the infant's gaze remained at the centre of the screen or the infant's first fixation was on an area of the screen that was not on a face, a score of 0.5 was given, based on the assumption that these fixation behaviours did not indicate discrimination between the stimuli. Correct performance was thereby calculated as proportion of first fixations made to the upright face.
A diagnostic criterion for crowding is that it rarely occurs in the fovea (Pelli et al., 2004
). To confirm the lack of a crowding effect during foveal viewing, we separately calculated an upright face preference score, indexing the proportion of time the infant spent fixating the upright face. Preferential looking to the upright face was computed for each trial by dividing the time spent fixating the upright face by the total time spent fixating both faces, ranging from 0 (never looked at upright face) to 1 (only looked at upright face), with 0.5 considered the chance level. Upright face preference scores were equivalent at the three eccentricities so the average score was used as a measure of foveal (0°), or free-viewing, performance on each trial. Because the infant’s gaze was directed to the upright face, the image must therefore have been projected onto the fovea. This measure allowed us to verify both that infants with fragile X syndrome do exhibit a face inversion effect akin to neurotypical infants and that foveal performance did not differ between uncrowded and crowded face conditions for each group of infants.
A logistic function was fit to each infant's data as a function of eccentricity and crowding condition using the psignifit toolbox for MATLAB (version 2.5.6), which implements the maximum-likelihood procedure described by Wichmann and Hill (2001)
. For all participants, the upper asymptote of the function was fixed at 1 (ceiling performance) and the lower limit was set to 0.5 (chance performance). To estimate parameters, threshold, slope and error, a bootstrapping technique was used which included 5000 replications for each fitted function. The criterion for including infants in the analyses was the goodness of fit of the function, evaluated using deviance scores >2 SDs above the mean of the group of infants (Wichmann and Hill, 2001
). Two infants with fragile X syndrome were excluded from the analyses. Threshold was defined as the eccentricity value yielding upright face discrimination performance of 0.75. Higher threshold values indicate greater spatial resolution of attention.
There was no difference in the number of trials successfully completed by infants in each group [neurotypical: mean = 50.34, SD = 19.57; fragile X syndrome: mean = 44.00, SD = 13.23; F(1, 63) = 2.31, P = 0.134]. Additionally, there was no difference in performance between male and female infants within each group, so this variable was removed from further analyses.
A repeated measures mixed-model ANOVA with eccentricity (0, 3, 6, 10°) and crowding (uncrowded or crowded faces) as within-subject factors and group (neurotypical or fragile X syndrome) as a between-subject factor was conducted on performance. Unless otherwise noted, all P-values reported are Bonferroni corrected for multiple comparisons. A significant main effect of eccentricity [F(3, 60) = 89.95, P = 0.0001, ηp2 = 0.818] and significant interaction effects between eccentricity and crowding [F(3, 60) = 4.53, P = 0.006, ηp2 = 0.185] and eccentricity and group [F(3, 60) = 3.16, P = 0.031, ηp2 = 0.137] were identified. The same ANOVA with the foveal (0°) eccentricity condition removed verified the presence of a significant main effect of eccentricity [F(2, 61) = 82.89, P = 0.0001, ηp2 = 0.731] and a significant interaction effect between eccentricity and crowding [F(2, 61) = 6.294, P = 0.003, ηp2 = 0.171], confirming that the crowding effect did not differ with or without the upright face preference score. These results revealed that infants’ ability to discriminate the upright Mooney face in the periphery decreased as a function of eccentricity, and was significantly worse when the faces were crowded (A). The interaction effect between eccentricity and group was driven by higher foveal performance in infants with fragile X syndrome compared with neurotypical infants in both the uncrowded and crowded conditions. Contrary to our prediction, no significant group difference was found in either analysis, indicating that crowding did not differ between infants with and without fragile X syndrome.
Figure 2 (A) Mean upright face preference score (±SEM) as a function of eccentricity for uncrowded and crowded faces, by group; (B) Mean upright face discrimination threshold (±SEM) for uncrowded and crowded faces, by group. Higher thresholds signify (more ...)
To further examine the effect of flankers on upright face preference for each group of infants, we conducted within-group paired-samples t-tests (two-tailed) at each eccentricity. Both groups of infants showed a significant difference between performance in the uncrowded and the crowded conditions when faces were presented at 3° [neurotypical: t(31) = 2.14, P = 0.040; fragile X syndrome: t(31) = 4.31, P = 0.0001], but not at further eccentric locations. At 6 and 10°, infants’ performance was not different from change in either the uncrowded or crowded conditions. Critically, flankers did not impair visual preference for the upright face when viewed foveally, consistent with the definition of crowding and distinguishing it from a masking process, which would have prevented both detection and discrimination independent of eccentricity.
Individual infant upright face discrimination thresholds from psychometric function fits were analysed using a repeated measures mixed-model ANOVA with crowding (uncrowded or crowded faces) as the within-subject factor and group (neurotypical or fragile X syndrome) as the between-subject factor, which yielded a significant main effect of crowding [F (1, 62) = 72.55, P = 0.0001, ηp2 = 0.539], reflecting higher eccentricity thresholds (better performance) when the faces were uncrowded. No other effects were found. Pair-wise t-tests (two-tailed) confirmed significantly higher sensitivity in the uncrowded (neurotypical: mean = 3.79, SD = 2.48; fragile X syndrome: mean = 3.89, SD = 1.09) compared with the crowded (neurotypical: mean = 1.97, SD = 0.97; fragile X syndrome: mean = 2.24, SD = 0.79) condition in both groups of infants [neurotypical: t(31) = 5.25, P = 0.0001; fragile X syndrome: t(31) = 7.73, P = 0.0001]. Therefore, flankers impacted upright face discrimination equally in infants with and without fragile X syndrome (B).
To compare the effect of crowding between individual infants, a threshold difference score was calculated by subtracting the threshold obtained for crowded faces from the threshold obtained for uncrowded faces. Difference scores were significantly positive for both groups [neurotypical: t(31) = 2.17, P = 0.038; fragile X syndrome: t(31) = 4.31, P = 0.0001], reflecting higher eccentricity limits for uncrowded faces. This analysis verifies that crowding-specific processes reduced upright face discrimination performance and that the decrement in discrimination did not differ between groups.
We also examined the relationship between uncrowded and crowded thresholds and other variables of interest. For neurotypical controls, chronological age was positively correlated with uncrowded threshold values [r(32) = 0.384, P = 0.030] such that older infants exhibited greater eccentricity limits for discrimination of the upright face presented in isolation. For infants with fragile X syndrome, there was no significant relationship between age and either threshold measure. Additionally, no significant relationship was found between molecular variables (CGG repeat length and FMR1 messenger RNA levels) and spatial thresholds in infants with fragile X syndrome.
The goal of this experiment was to use visual crowding to quantify the spatial resolution of visual attention in infants with fragile X syndrome compared with neurotypical infants. The psychophysical measure of upright face discrimination thresholds established that infants with fragile X syndrome were able to identify the uncrowded Mooney face to a limited extent in the periphery and that crowding interfered with identification at 3°. Thresholds for uncrowded and crowded face discrimination in infants with fragile X syndrome were equivalent to those measured in developmental age-matched neurotypical infants, suggesting that spatial resolution is intact in infants with fragile X syndrome.
The use of Mooney faces as stimuli had the advantage of allowing us to examine several aspects of face processing abilities in infants with fragile X syndrome. This is the first study to examine face detection in infants with fragile X syndrome. We found that infants with fragile X syndrome were able to perceive Mooney faces, as illustrated by a selective visual preference for the upright relative to the inverted face. Since Mooney faces lack individual facial features and cannot be recognized by bottom-up processes, a preference for the upright face suggests that, at the ages tested, individuals with fragile X syndrome possess an intact holistic face processing system. Holistic face processing is thought to rely on analysis of the configuration or relations between features, and therefore demonstrates Gestalt processing of the face as a whole unit (Tanaka and Farah, 1993
). Further, these results, for the first time, establish the presence of the face inversion effect (Yin, 1969
) in individuals with fragile X syndrome. Lastly, given that infants with fragile X syndrome showed no gaze aversion to the faces, these data suggest that avoidance of social stimuli may develop after infancy, perhaps as a coping mechanism in response to increased social demands.
Crowding is believed to impose the fundamental bottleneck preventing object-level visual information from reaching conscious awareness (Levi, 2008
; Pelli and Tillman, 2008
; Whitney and Levi, 2011
), thereby providing a quantitative measure of the resolution of spatial attention (Intriligator and Cavanagh, 2001
) and allowing us to study the development of spatial attention in infants with fragile X syndrome. The current findings revealed no difference in the resolution of spatial attention between infants with and without fragile X syndrome. One explanation for our findings is that spatial attention in isolation may be relatively unaffected in infants with fragile X syndrome, but that a deficit lies in the integration of spatial and temporal attention. Previous reports of visual processing impairments in individuals with fragile X syndrome utilized tasks that did not separate these two types of attention, and therefore required simultaneous attention to both static and dynamic information. The crowding task used in this experiment aimed to identify the limit of spatial attention for the recognition of static faces in the peripheral visual field, eliminating temporal attention. It is also plausible that the spatial resolution of attention required for visual perception is different from the spatial resolution required for visually guided actions, such as pointing or grasping (Bulakowski et al., 2009
), and that the latter is selectively impacted in individuals with fragile X syndrome. Accordingly, identification of a target in a visual search task would be more impacted by the clutter, or crowding, caused by distractors, as was found in toddlers with fragile X syndrome (Scerif et al., 2004
). These two explanations need not be mutually exclusive, as spatial and temporal information must be combined for accurate visuomotor responses.
Although chronological age predicted uncrowded thresholds in neurotypical controls, it did not do so for infants with fragile X syndrome. This may be the result of a measurable increase in peripheral visual acuity, particularly for medium to high spatial frequencies, in the younger aged neurotypical infants, but less so in the older age range of the infants with fragile X syndrome. This reasoning is in line with evidence that processing of fine detail is dependent on a sensitive window of postnatal visual experience (Maurer and Lewis, 1993
; Dobkins et al., 1999
; Birch and O'Connor, 2001
Infants in both groups spent a greater proportion of time fixating the upright face, whether uncrowded or crowded. The finding that infants with fragile X syndrome spent an even greater proportion of time fixating the upright face relative to neurotypical controls was unexpected and may be accounted for by delayed disengagement from the upright face. At eccentricities beyond the fovea, performance in both the uncrowded and crowded conditions eventually dropped to chance levels, which is expected for peripheral face recognition (Mäkelä et al., 1993
; McKone, 2004
). This decline in performance is likely the result of both reduced acuity and within-face crowding in the periphery (Martelli et al., 2005
; Farzin et al., 2009
). Nevertheless, because infants were able to direct their first fixation precisely to either the upright or inverted face as far as 10° eccentricity, we are certain that infants were able to ‘perceive’ the stimuli in their periphery and it is therefore unlikely that acuity limits alone explain the decline in uncrowded performance.
At this point, the resolution of spatial visual attention in older children, adolescents and adults with fragile X syndrome remains unknown. Hence, although the limits of resolution are comparable between infants with and without fragile X syndrome early in development, we do not know if attentional resolution reaches an asymptotic level in individuals with fragile X syndrome that is lower compared with that of developmental age-matched neurotypical individuals. Literature on the maturation of spatial visual attention in neurotypical individuals is also scarce so it is yet to be determined at what age adult levels of resolution are attained. In any case, the spatial resolution of infant attention is comparable in those with and without fragile X syndrome.