Minimally verbal children with autism may ‘see the point, but do not (always) point to what they see’

Minimally verbal children with autism may ‘see the point, but do not (always) point to what they see’ | bioRxiv /* */ /* */ <!-- <!-- /*! * yepnope1.5.4 * (c) WTFPL, GPLv2 */ (function(a,b,c){function d(a){return"[object Function]"==o.call(a)}function e(a){return"string"==typeof a}function f(){}function g(a){return!a||"loaded"==a||"complete"==a||"uninitialized"==a}function h(){var a=p.shift();q=1,a?a.t?m(function(){("c"==a.t?B.injectCss:B.injectJs)(a.s,0,a.a,a.x,a.e,1)},0):(a(),h()):q=0}function i(a,c,d,e,f,i,j){function k(b){if(!o&&g(l.readyState)&&(u.r=o=1,!q&&h(),l.onload=l.onreadystatechange=null,b)){"img"!=a&&m(function(){t.removeChild(l)},50);for(var d in y[c])y[c].hasOwnProperty(d)&&y[c][d].onload()}}var j=j||B.errorTimeout,l=b.createElement(a),o=0,r=0,u={t:d,s:c,e:f,a:i,x:j};1===y[c]&&(r=1,y[c]=[]),"object"==a?l.data=c:(l.src=c,l.type=a),l.width=l.height="0",l.onerror=l.onload=l.onreadystatechange=function(){k.call(this,r)},p.splice(e,0,u),"img"!=a&&(r||2===y[c]?(t.insertBefore(l,s?null:n),m(k,j)):y[c].push(l))}function j(a,b,c,d,f){return q=0,b=b||"j",e(a)?i("c"==b?v:u,a,b,this.i++,c,d,f):(p.splice(this.i++,0,a),1==p.length&&h()),this}function k(){var a=B;return a.loader={load:j,i:0},a}var l=b.documentElement,m=a.setTimeout,n=b.getElementsByTagName("script")[0],o={}.toString,p=[],q=0,r="MozAppearance"in l.style,s=r&&!!b.createRange().compareNode,t=s?l:n.parentNode,l=a.opera&&"[object Opera]"==o.call(a.opera),l=!!b.attachEvent&&!l,u=r?"object":l?"script":"img",v=l?"script":u,w=Array.isArray||function(a){return"[object Array]"==o.call(a)},x=[],y={},z={timeout:function(a,b){return b.length&&(a.timeout=b[0]),a}},A,B;B=function(a){function b(a){var a=a.split("!"),b=x.length,c=a.pop(),d=a.length,c={url:c,origUrl:c,prefixes:a},e,f,g;for(f=0;f<d;f++)g=a[f].split("="),(e=z[g.shift()])&&(c=e(c,g));for(f=0;f<b;f++)c=x[f](c);return c}function g(a,e,f,g,h){var i=b(a),j=i.autoCallback;i.url.split(".").pop().split("?").shift(),i.bypass||(e&&(e=d(e)?e:e[a]||e[g]||e[a.split("/").pop().split("?")[0]]),i.instead?i.instead(a,e,f,g,h):(y[i.url]?i.noexec=!0:y[i.url]=1,f.load(i.url,i.forceCSS||!i.forceJS&&"css"==i.url.split(".").pop().split("?").shift()?"c":c,i.noexec,i.attrs,i.timeout),(d(e)||d(j))&&f.load(function(){k(),e&&e(i.origUrl,h,g),j&&j(i.origUrl,h,g),y[i.url]=2})))}function h(a,b){function c(a,c){if(a){if(e(a))c||(j=function(){var a=[].slice.call(arguments);k.apply(this,a),l()}),g(a,j,b,0,h);else if(Object(a)===a)for(n in m=function(){var b=0,c;for(c in a)a.hasOwnProperty(c)&&b++;return b}(),a)a.hasOwnProperty(n)&&(!c&&!--m&&(d(j)?j=function(){var a=[].slice.call(arguments);k.apply(this,a),l()}:j[n]=function(a){return function(){var b=[].slice.call(arguments);a&&a.apply(this,b),l()}}(k[n])),g(a[n],j,b,n,h))}else!c&&l()}var h=!!a.test,i=a.load||a.both,j=a.callback||f,k=j,l=a.complete||f,m,n;c(h?a.yep:a.nope,!!i),i&&c(i)}var i,j,l=this.yepnope.loader;if(e(a))g(a,0,l,0);else if(w(a))for(i=0;i (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0];var j=d.createElement(s);var dl=l!='dataLayer'?'&l='+l:'';j.src='//www.googletagmanager.com/gtm.js?id='+i+dl;j.type='text/javascript';j.async=true;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-M677548'); Skip to main content Home About Submit ALERTS / RSS Search for this keyword Advanced Search New Results Minimally verbal children with autism may ‘see the point, but do not (always) point to what they see’ View ORCID Profile Hannah S. Sykes-Haas , View ORCID Profile Yoram S. Bonneh doi: https://doi.org/10.1101/2025.06.26.661808 Hannah S. Sykes-Haas 1 School of Optometry and Vision Science, Bar-Ilan University , Ramat Gan, Israel 2 Multidisciplinary Brain Research Center, Bar-Ilan University , Ramat Gan, Israel 3 Shtilim special education pre-school for children with autism , Jerusalem, Israel Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Hannah S. Sykes-Haas For correspondence: hsykeshaas{at}gmail.com yoram.bonneh{at}gmail.com Yoram S. Bonneh 1 School of Optometry and Vision Science, Bar-Ilan University , Ramat Gan, Israel 2 Multidisciplinary Brain Research Center, Bar-Ilan University , Ramat Gan, Israel Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Yoram S. Bonneh For correspondence: hsykeshaas{at}gmail.com yoram.bonneh{at}gmail.com Abstract Full Text Info/History Metrics Supplementary material Preview PDF Abstract During typical development, non-social visual object recognition emerges in the first year of life, engaging both low-level cues (e.g., color, orientation) and higher-level mechanisms involving inference and prior knowledge. Little is known about how these processes function in minimally verbal children with autism (mvASD). We studied 22 children with mvASD using touchscreen-based oddball and contour detection tasks, targeting low-level (e.g., shape, orientation) and mid-level (e.g., Kanizsa figures, 3D shapes) visual stimuli, measuring both pointing and eye-gaze responses. All children detected the oddball in the easiest condition with faint distractors, and approximately half succeeded across all low-level tasks. Notably, some “high performers” showed reduced accuracy under mid-level conditions with greater stimulus complexity. Strikingly — and not originally anticipated — several “low performers” who failed to point correctly nonetheless fixated on the correct target. In the Kanizsa oddball task, several mvASD participants, unlike typically developing (TD) peers, consistently pointed to local inducers rather than to the center of the illusory triangle. While the overall deterioration in performance with increased visual complexity suggests that mvASD visual perception may rely on low-level representations with attenuated inference-based processing, the dissociation between gaze and pointing — along with atypical local-pointing behavior — indicates that performance depends not only on what is perceived, but also on how they use the visual signal to drive their behavior. They may, quite literally, see the point — but not point to what they see. Lay Summary Minimally verbal children with autism are often treated as a uniform group, but this study revealed wide differences in how they process and respond to visual information. While many accurately perceived simple images, some struggled with more complex ones—and several looked directly at the correct target but pointed elsewhere, revealing a disconnect between perception and action. These findings highlight the need to consider both individual differences and the role of visual complexity and layout, as well as the importance of assessment methods that go beyond motor (pointing) responses to better understand and support individuals with mvASD. Introduction AG, a beautiful six-year-old non-verbal girl with autism, still had no alternative augmentative communication (AAC) system in place. Despite intensive efforts, she struggled to find the image of a desired item even when only choosing between two. Inspired by a study in a “supposedly” unrelated field of visual crowding (Manassi et al ., 2013), we modified the visual presentation of her cards by presenting the target card alongside seven black cards. AG immediately selected the correct image, and she learned to discriminate desired items from distractor images as well. AG appears to have been challenged by visual presentation and processing , rather than semantic knowledge and the communicative nature of the AAC tool (first author HS, clinical case description) . During the first year of life, basic, non-social, visual object recognition skills emerge in typically developing children ( Nishimura et al., 2009 , McKyton et al., 2015 , Haaf et al, 2003 , Nishihara et al., 2009). These foundational abilities involve both low-level visual cues (e.g. color and orientation) and more advanced and complex mechanisms that rely on prior knowledge, integrative and inference-based processing (e.g. shape from shading) ( Manassi et al., 2013 , McKyton et al, 2015 and Bertone et al., 2005 ). In autism these early visual processing mechanisms may diverge from typical developmental pathways (e.g. Glila et al, 2015 and Ostrolenk et al. 2017 ). In a seminal study by Bertone et al. (2005) they demonstrate that autistic individuals are superior at perceiving static simple, luminance-defined gratings (first-order stimuli) but show diminished performance with more complex, texture-defined gratings (second-order stimuli) requiring integrative and inference-based visual processing. They propose the ‘complexity-specific hypothesis’ of autism ( Bertone et al., 2005 , Bertone et al., 2003 ). In the ‘Enhanced Perceptual Functioning’ (EPF) theory, Mottron and colleagues further propose that local low-level processing superiority may account for positive symptoms in autism, such as detail-focused perception, whereas reduced capacity for processing complex material could contribute to negative symptoms (e.g. Mottron et al. 2006 and Kéita et al., 2014). In general, an extensive body of research exists examining visual perception and cognition in verbal individuals with autism, indeed supporting inter alia superior local (simple) processing and some suggesting diminished global (complex) processing ( Dakin and Frith, 2005 , Frith and Happe, 1994, Booth and Happe, 2018, Simmons et al., 2009 , Behrman et al., 2006). Despite these findings and despite the widespread reliance on visual aids ( Arthur-Kelly et al., 2009 , Hayes et al., 2010 and as shown in the clinical example above), relatively little is known about how these perceptual patterns manifest in minimally verbal individuals with autism (mvASD), a subgroup representing the severe end of the autism spectrum and accounts for 25-30 % of the autism population (Hus Bal et al., 2016 , Tager-Flusberg and Kasari, 2013 , Rose et al., 2016 ). Given the ‘complexity-specific hypothesis’ and EPF, and how these putative differences in (visual) processing mechanisms may account for positive as well as negative symptoms in autism it seems pertinent to explore the most severely affected and whether differences in the ability to process simple versus more complex low level visual stimuli are similarly evident—or perhaps even more pronounced—among mvASD individuals. Furthermore, investigating low-level visual processing in mvASD children is a critical starting point, as it addresses the fundamental mechanisms that emerge early in development and provide a foundation for higher-order cognitive and perceptual functions. Whilst there has recently been an upsurge in mvASD research ( Courchesne et al., 2015 , Girard et al., 2023 , Brignell et al., 2018 , Hus Bal et al., 2016 , Posar and Visconti, 2020 and more) very few have focused on visual perceptual processing within this clinical subgroup. To our knowledge, investigations into basic low level visual processing among the minimally verbal with autism remains an un(der)explored area. Notably, Soulières and colleagues ( Courchesne et al. 2015 and Girard et al., 2023 ) find typical performance on visual search (VS) and children embedded figures tasks (CEFT) among some children with mvASD, further confirming local visual perceptual strengths also among mvASD individuals. It should be noted that tasks were conducted in board form with removable pieces, rather than reporting by pointing. These more tangible assessment formats yield outcomes that may indicate that performance is not only dependent on visual perception, but also on how presentation mode (i.e. inter alia visual and motor task requirements) modulates behavior and performance, providing better insights into the challenges and abilities of the individual with mvASD. In the current study, we investigated early visual processing in mvASD via pointing and eye-gaze. We drew inspiration from a developmental study by McKyton et al., (2015) , which employed touch-screen-based oddball tasks to assess low- and mid-level vision (in children following late removal of congenital cataracts). Their paradigm allowed for modulations so that reliance on verbal instructions was removed, allowing participants to imitate modelled task skills during practice, enabled short, repeated trial runs and unlimited response time. This approach is well suited for mvASD populations, with their complex, unique and variable behavioral and cognitive profiles ( Kasari et al., 2013 ; Matsuzaki et al., 2019 ; Tager-Flusberg et al., 2017 , Bauminger-Zviely et al., 2020 ). We examined pointing and eye-gaze in mvASD children during a series of visual tasks (oddball and contour detection tasks) probing low- (e.g., color) and mid-level visual stimuli (e.g., 3D shapes and Kanizsas). Results revealed that mvASD children are not a homogeneous group and performance varied as complexity increased. In the Kanizsa oddball successful pointing was often guided by the local elements (i.e. the inducers), in contrast to TDs’ who point to the center of the Kanizsa. Strikingly, eye-gaze was superior to pointing among poorest performers. These findings suggest that mvASD children could visually locate targets even when their motor responses were less accurate. Methods Participants Experimental group Twenty-two school-aged children with minimally verbal autism (mvASD) participated in this study. Their age range was 6.9–11.1 years (M = 8.9 ± 1.3 years) and included 16 males and 6 females. The children were recruited from three special-education schools and pre-schools in Israel with a relatively large cohort of students with mvASD. All mvASD participants met inclusion criteria for the autism group. Each child had received a formal clinical ASD diagnosis according to DSM-IV-TR (APA, 2000) or DSM-5 (APA, 2013) criteria by two independent accredited autism diagnosticians, as required by Israeli Health Ministry and Ministry of Education for entry into the special education school system for students with autism. At least one diagnostician was a physician (child neurologist, psychiatrist, or developmental pediatrician), and the second was either a physician or developmental/clinical child psychologist. None were affiliated with the study. The ASD diagnosis was further confirmed through teacher-reported scores above the autism cutoff (≥15) on the Social Communication Questionnaire (SCQ; Rutter et al., 2003 ). mvASD criterion Participants met the criterion for minimal spoken language if caregivers (parent and teachers) indicated that they either rarely or never used meaningful phrase-level speech spontaneously in everyday situations, or if their functional spoken vocabulary consisted of roughly 30 or fewer distinct communicative words or phrases, not including echoed language ( Plesa Skwerer et al., 2016 ). Ravens Matrices MvASD participants underwent assessments of fluent reasoning using Raven’s Colored Progressive Matrices in both booklet and puzzle board form (RCPM; Raven et al., 1998 ). This RCPM is believed to be a core measure of fluid intelligence, assessing abilities such as rule inference, goal management, and abstract reasoning independent of language ability ( Dawson et al., 2007 ; see Table 1 for summary of descriptive measures and appendix 1 for descriptive measures for individual participants). For three mvASD participants (Pseudonyms Levi, Lia, and Haim), RCPM performance data is missing due to leaving the school or behavioral challenges (see Appendix 1). These cases were excluded from analyses involving RCPM but retained for all other analyses where applicable. Missing data were not imputed. View this table: View inline View popup Download powerpoint Table 1. Summary of descriptive measures 1 Low-Verbal Investigatory Screener 4.0 Parents completed the Low-Verbal Investigatory Survey 4.0 (L-VIS 4.0; Naples et al., 2022, see also: https://autismlanguagelab.psychology.uconn.edu/low-verbal-investigatory-survey-elvis-project-home ), a newly developed brief (5-minute) caregiver-report tool assessing communication in minimally verbal autistic children. In this study, L-VIS 4.0 was used solely to assess communication abilities. Although the original factor analysis supports a two-factor model, only the first factor—Communication Abilities—was statistically validated and analyzed. This factor reflects typical verbal and receptive communication, with high internal consistency and strong correlations with standardized measures (Vineland-II: 0.73; PLS-5: 0.61; Stanford-Binet non-verbal: 0.53). The second factor, Atypical Communication, was not used for composite scoring due to mixed loadings and interpretive concerns. However, we report one item from this factor—production of unusual sounds (e.g., shrieks)—as it was a strong predictor of performance across multiple gold-standard assessments (PLS, VABS, and Stanford-Binet IQ). For three participants (See appendix 1) the L-VIS 4.0 was completed by trained care staff familiar with the child for over a year, due to lack of parental response. Exclusion criteria Participants were excluded if they had known comorbid developmental, neurological, or genetic conditions (e.g ADHD, CP, Angelman’s Syndrome, Icardy syndrome etc.), with the exception of medically controlled epilepsy, developmental delay and intellectual disability, as standardized tests and/or diagnostician’s clinical impression may at times underestimate cognitive abilities in mvASD ( Courchesne et al., 2015 , Bauminger-Zviely et al. 2020 , Pizzano et al 2024 ). Four participants did not complete all experimental tasks. One male mvASD participant was excluded from two oddball tasks because of behavioral challenges, while a female mvASD participant moved abroad before finishing the final task. Two other participants joined the study later and only participated in selected tasks. Sixteen children in the autism group were receiving psychotropic medications to manage behavioral challenges, maintaining focus and/or sleep regulation. They were prescribed Risperidone (Risperdal), Medical Cannabis, Carbamazepine (Tegretol) Ritalin, Atent and melatonin. All mvASD participants were given pseudonyms starting with the initials ‘H’ or ‘L’ for (High and Low performers, depending on their overall performance on the various tasks explained in the following). Control group A control group of twenty-three typically developing (TD) children participated in the study (age range: 6-13 years; M = 9 ± 2.1 years), comprising 11 males and 9 females. None of the control group participants had any known developmental, neurological, cognitive, or genetic conditions, and all had normal or corrected-to-normal vision according to parental report. Ten TDs completed all 8 conditions, 19 participated in at least 6, and 14 were tested on the Kanizsa oddball condition, which was added later. Two of these were tested only on this condition to increase the sample size. Ethics approval was obtained from the Chief Scientist of the Israeli Ministry of Education and the Ethics Committee at Bar-Ilan University. Informed consent was obtained from all parents (mvASD and TD), and data collection took place at either the child’s home or at their educational institution per parental preference. All TDs and one mvASD participant were tested in their homes. Apparatus Testing was conducted using a Lenovo Yoga C940 laptop featuring a 14-inch touchscreen display with a resolution of 1920 × 1080 pixels and a refresh rate of 60 Hz. Eye-tracking data was recorded using the Tobii Eye Tracker 4C, operating at a 90 Hz sampling rate and providing gaze tracking accuracy of 0.5-1 degrees. The Tobii eye-tracker was selected for its non-invasive nature, allowing participants freedom of movement while maintaining tracking accuracy. The experimental setup was stabilized using a Tobii EyeMobile computer tray mounted on an adjustable monitor arm. This configuration ensured consistent positioning and protected the devices from sudden unpredictable behaviors. Stimulus presentation was managed through the PSY platform, an in-house software for psychophysics and eye tracking developed by Y.S.B. (second author). Stimuli All tasks involved visual oddball or contour detection arrays presented on a touchscreen. The stimuli spanned a range of perceptual complexity, from basic features (e.g., luminance contrast, color hue) to mid-level (e.g., 3D shading, Kanizsa figures, contour integration), similar to McKyton et al., (2015) . Oddball stimuli were displayed in 2×3 and 2×2 grid configurations at fixed screen locations, with standardized viewing distance (∼50–60 cm), visual angles (∼2.5°–6.5°), and fixed inter-stimulus spacing. The stimuli are described in Figure 1 , showing the actual stimulus displays for the 2×3 configurations, with trials differing only in the position of the oddball. These displays were all ∼10 cm (center-to-center) in width, except for the displays in Figure 1 , b1 (∼17 cm), and b4 (∼13 cm). All the other sizes and positions were in proportion to the display size as shown. The background for most stimuli was white, gray, or black (luminance in cd/m 2 , a1: 1.5; a2-3: 75; a4, b4: 110; b1: 335; b2-3: 100). The patches in a3 were a white target, and black, dark grey, or mid grey distracters (145, 3, 31, 61 cd/m 2 respectively). The grating stimuli in a4 were square patches of 1 cpd sine-wave gratings with 32% contrast, vertical with horizontal oddball. The stimuli in b4 were a second-order version of a4, created by a sine-wave modulation of the contrast of a random dot pattern (1-pixel texels with random luminance) as in Bertone et al. (2005) . Download figure Open in new tab Figure 1. The experimental paradigms and sample stimuli (a) Low-level oddball paradigms. (a1) shapes ( a2 ) Color ( a3 ) Black/white saliency ( a4 ) 1 st orientation gratings, ( a5 ) white/black w/eye-gaze. (b) Mid-level oddball and contour paradigms., (b1) Kanizsa, (b2) Light from above, ( b3 ) 3D box orientation, ( b4 ) 2 nd order orientation gratings ( b5 ) Circular contours with increasing distractor density. The contour detection stimuli ( Figure 1 .b5) were generated as configurations of symmetric Gabor patches (wave length l and envelope s of 16 pixels) with 30% contrast on a gray (60 cd/m 2 ) background, similar to Bonneh & Sagi (1998 , but in which a circle is embedded in a random background ( Field et al., 1993 ). First, a background of randomly oriented Gabor patches arranged on a grid with a specific spacing that varied across conditions was set to cover 1900×600 pixels rectangle, with a position jitter radius per patch of 48 pixels. Then, a circle of patches with a radius of 160 pixels, inter-patch interval of 88 pixels (5.5 l), and uniform tangential orientation jitter of +/-20 deg, was added in one of 4 positions (+/-450 and +/-250 pixels for x and y respectively). Background patches that were closer than 48 pixels to a target patch (center-to-center) were erased. There were 3 contour difficulty levels in separate runs according to the background inter-patch interval: Easy (640, 192 pixels), Medium (128, 96 pixels), and Hard (64,48 pixels), two intervals mixed per level. Full stimulus parameters are provided in Table S1 of the Supplementary material. Procedure Testing was conducted in a quiet room (home or school) with lighting at ∼500 lux. The experimenter (first author, HSH), an experienced clinician in mvASD, ran all sessions. Participants sat ∼60 cm from the screen and could move freely. The experimenter sat to their right, with rewards placed 70–80 cm away. Each participant completed 8–10 sessions over 3–8 months. Sessions lasted 10–15 minutes, with breaks based on child’s need and motivation. Participants could leave at any time. Sessions were scheduled no more than once weekly, with longer gaps due to holidays or absences. Before each block, participants received brief verbal instructions. For oddball tasks, they were told to “find/point to/look at the one that’s different”; for the circular contour task, “find/point to the circle.” There were 7 oddball pointing conditions 3 circular contour detection by pointing conditions, and one oddball task requiring gaze or/and pointing ( Figure 1a and 1b ). To address language limitations, each condition began with 8–10 practice trials. The experimenter modelled pointed to the correct item and after 3–4 failed modelled practice trials she would use hand-over-hand guidance (if child consented). Each condition had 10–32 trials, with oddball targets randomly positioned in 2×2 or 2×3 grids at any of the 4 or 6 possible locations. Conditions were kept brief (10–30 trials, ∼20s–2min) to maintain engagement and allow breaks. Participants gradually tolerated more trials as they acclimated to test setting. Trial counts per run were: Color, shapes, BWG: 4 runs × 10 trials; First-/Second-order: 2 runs × 20 trials; Box, light-from-above: 4 runs × 18 trials; Kanizsa: 1–2 runs × 24 trials; Contour detection: 3 difficulty levels, 16 trials each (4 locations of contour × 2); tested once or twice per participant (48–96 trials total). Actual trial numbers varied due to performance or trial rejection (e.g., during double-clicks second trial was discarded, see below). In the black/white eye-gaze oddball task, each run included 12 trials. Participants completed 3–16 runs (e.g., “Loni” = 3; “Lara” = 16; 8 participants = 5–10). Some trials were unusable due to movement, looking away, or eye-tracker obstruction. Only gaze data within the stimulus region were analyzed. Three participants contributed 7– 8 valid trials; five, 16–26 trials; two, ∼75 trials. In pointing tasks, new trials began upon screen touch. Responses <200 ms after previous touch (double clicks) were discarded. In eye-gaze-only tasks, displays remained for 2.5 seconds. Eye-tracking used 3-point calibration. Rewards (e.g. food, toys, videos) were tailored based on caregiver input and provided for task success and cooperation. Data Analysis Pointing Oddball tasks Correct responses defined as touches within a 150-pixel radius of the target oddball. In Kanizsa oddball trials, average distance from the target center was also calculated. Contour detection Correct responses defined within 200 pixels of the target center. To ensure validity, only touches occurring >200 ms after the previous touch were included. Eye-Gaze We applied two separate criteria for ‘correct’ eye-gaze fixations: (1) Momentary fixation accuracy: Fixation within a 200-pixel radius of the target center in the 300–700 ms window post-trial onset; and (2) Sustained/competitive fixation accuracy: Cumulative fixation time on target > cumulative fixation on any distractor, within the range of 0.4-1.5s from stimulus onset. For both measures, eye-gaze data were only included if gaze fell within a 400×500-pixel region at the screen center. The 200-pixel radius criterion for eye-gaze was adopted due to suspected compromised prior calibration and fixation precision due to excessive movement, with precision differences favoring pointing. These adjustments align with known autism gaze difficulties and low calibration accuracy (e.g. Wass et al., 2015 ). While this liberal approach may slightly reduce the precision of gaze data, it ensures a fair comparison across groups and accommodates the practical constraints of testing mvASD participants. Reaction Times (RTs) RTs were recorded throughout the study. All responses >200 ms were included in accuracy analyses. For RT analyses (including visualizations of median RTs across mvASD subgroups and TD controls), only correct responses with RTs <3000 ms were considered, to exclude trials with extended behavioral delays while retaining valid responses. Such delays often occurred among mvASD participants, as some children intermittently took breaks—such as standing up or walking away—before returning to complete a trial. RCPM Scoring RCPM raw scores were converted to age-normed percentiles ( Raven et al., 1998 ), interpolated linearly, and scaled to approximate IQ scores (IQ = 100 + z×15), as in Courchesne et al. (2015) . Subgrouping Initial data exploration (scatterplots, performance patterns) indicated distinct performance subgroups. K-means clustering (k=2 and k=3, final k=2) was used to define ‘high’ and ‘low’ performance groups. Participants were labelled using pseudonyms, with initials L for “low” performers (e.g., “Lia”) and H for “high” performers (e.g., “Hugo”), based on performance during the ‘shape’ oddball condition. Statistical Analyses Paired-sample t-tests evaluated performance on low- and mid-level visual tasks within the mvASD group. Independent sample t-tests compared performance between ‘high’ and ‘low’ performers and Kanizsa off-center pointing performance between mvASD and TD participants. A two-way ANOVA assessed the effects of group (high vs. low) and task level (low vs. mid), and their interaction. Additional paired-sample t-tests compared eye-gaze vs. pointing performance in the black/white salience oddball task (low performers only). In comparing levels as well as eye-gaze and pointing, we also computed Effect Size (Cohen’s D) as well as AUC (The Area Under the ROC Curve). Results Oddball and contour detection tasks In the oddball and contour detection tasks participants were required to detect visual stimuli among distractors. Performance was measured as the proportion correct (P-correct) for each mvASD participant across eight different visual stimuli. Figure 2 illustrates the individual performance accuracy (p-correct) for each mvASD in each condition, displayed in ascending order from lowest to highest performance. Each bar represents a single participant. In stimulus tasks with two or more conditions, ordering was sorted according to one of the conditions (e.g. shapes task according to performance on square oddball). The results show large variability in accuracy among participants, with some showing near chance level performance and others at or near ceiling accuracy rates. The overall accuracy varied across stimuli , with higher performance observed for low level visual stimuli (Se Figure 1 ). We divided the mvASD participants into ‘low’ and ‘high’ performance subgroups according to a k-means clustering analysis (k = 2, see Methods), and used pseudonyms for each participant with initials “L” for low and “H” for high performers. Notably, most performers remained within their category of ‘low’ or ‘high’ throughout all 8 tasks. Typically developing controls (TDs) performed at or near ceiling (oddball p-correct average of 0.98-1, contour 0.95-1, results not shown). Overall, the mvASD group showed lower performance than controls, possibly reflecting attenuated visual processing, with subgroups observed within the mvASD group. Download figure Open in new tab Figure 2. Performance on the oddball and contour detection tasks via pointing (a) Black/white saliency, (b) color, (c) shapes (d) 1 st and 2 nd order orientation gratings, (e) Kanizsa, (f) light from above and (g) 3D box orientations and (h) circular contours among increasing no. of Gabor distractor elements (see f igure 1 for visual stimulus). Data were plotted with a bar for each mvASD participant sorted from low to high performance in one stimulus condition in each experiment (e.g. Shape experiment sorted according to square oddball condition). Pseudonyms were used for each participant grouped according to their performance on Shapes experiment (a) with initials “L” for low and “H” for high performance. Error bars denote 1SE of the mean across trials. Performance across visual stimulus levels To further investigate performance across different levels of visual stimuli, we conducted two analyses. In the first analysis, performance was compared between ‘assumed’ low- and mid-level visual stimuli, categorized as per McKyton et al. (2015) as well as our categorization of added stimuli in the current study, excluding easy and medium difficulty contours (see Methods, Stimuli’). As predicted, the results revealed a significant difference between the two conditions (t(19) = 4.47, p < 0.0005), though the area under the ROC curve (AUC) was relatively low (0.65), suggesting limited differentiation between levels ( Figure 3a ). Download figure Open in new tab Figure 3. Performance comparison during low and all mid-level visual tasks (a) Bee-swarm plot illustrating mvASD participants’ performance in assumed/pre-categorized low- and mid-level visual tasks. Each pair of circles connected by a gray line represents one participant. (b) Bee-swarm plot showing mvASD participants’ performance in performance-based/post-categorized low- and mid-level visual tasks, following reclassification based on observed performance values. (c) Diagonal scatterplot where each circle represents one mvASD participant’s performance across performance-based visual levels. Colored circles indicate clusters identified using K-means clustering (K = 3). For K = 2, clusters with near-ceiling low-level visual performance (orange and green circles) were merged into a single group. These merged clusters were used to define the “low” and “high” performance groups. To refine the analysis, stimuli were re-categorized based on observed performance including all tasks at or near ceiling (above 85% performance) as low-level, resulting in two mid-level stimuli being reclassified as low-level (Kanizsa and light from above). This adjusted categorization yielded a higher AUC score (0.74), reflecting improved differentiation (t(20) = 5.57, p < 0.0005). Findings suggest mvASD participants performed significantly better on low-level visual stimuli and performance-based post categorization suggests that additional (visual) variables, not captured by the original categorization, may influence visual processing and performance ( Figure 3b ). A diagonal scatterplot was used to depict the relationship between performance on low- and mid-level visual stimuli ( Figure 3c ) in accordance with performance-based recategorization of visual processing levels. Points clustering along the diagonal indicate similar performance across conditions, while points below the diagonal suggest higher accuracy in the low-level stimulus condition. More than 50% of mvASD participants reached ceiling performance on low-level stimuli, represented by points near the upper end of the low-level x-axis. In contrast, fewer than 25% of participants achieved ceiling performance in mid-level conditions, as shown by the spread of points along the mid-level y-axis. mvASD subgrouping Participants appeared to consistently fall into at least 2 subgroups. We performed k-means clustering analysis (for k = 2 and k=3) on the data, identifying two and three distinct clusters of participants, respectively, based on their performance ( Figure 3b ). Cluster 1 showed the highest accuracy across all stimuli, while cluster 3 displayed the lowest accuracy, with cluster 2 performing better on low level visual stimuli. In k=2 cluster 1 and 2 combined represent the “high” performance group ( Figure 3c ). An independent sample t-test comparing performance of these two sub-groups across all visual stimuli levels was highly significant (t(19) = 12.39, p < 0.0005), as shown in figure 4a . Further analyses comparing performance on low-level and mid-level visual stimuli separately within these groups also yielded highly significant differences (t(19) = 12.73, p < 0.0005 and t(19) = 5.93, p < 0.0005), respectively. Download figure Open in new tab Figure 4. Group (high, low and TD) comparisons, with tasks categorized according to performance-based/post-categorized low- and mid-level visual tasks (a) accuracy in all low and mid-level visual tasks; (b) accuracy in low-level visual tasks, (c) accuracy in midlevel visual tasks presented in bee-swarm charts. Each circle represents one mvASD participant, (d) all low and mid-level visual tasks among ‘low’ (red bars) and ‘high’ (blue bars) performers in bar graph and (e) median RT for correct responses in each stimulus condition among low and high performers and TDs. Note that difference in performance is highly significant among ‘high’ and ‘low’ performers (p<0.0005***) in all three comparisons. However, some ‘high’ performers deteriorate during mid-level tasks and performers group begin to overlap. Also, RT among high performers followed a similar path to TDs. MvASD is not a homogeneous clinical subgroup . To validate the subgroup differences identified through k-means clustering and t-tests, we conducted a two-way ANOVA with Group (low vs. high performers) and visual task Level (low vs. mid-level stimuli) as factors. Results confirmed a significant main effect of Group, F (1, 232) = 506.48, p < .001, and visual task Level, F (13, 232) = 7.42, p < .001. A significant interaction was also found, F (13, 232) = 1.86, p = .036, indicating that the performance gap between groups varied by stimulus complexity. Taken together, these findings support the t-test results illustrated in the b-swarm charts ( Figures 4b and 4c ) of consistent differences in performance between ‘high’ and ‘low’ performers across varying levels of visual complexity and confirm sub-grouping within mvASD. Findings further highlight that mid-level visual tasks pose particular difficulty for low-performing mvASD participants ( Figures 4c and 4d ) and that performance among some “high” performers deteriorate during mid-level visual stimuli ( Figure 4c ). Finally, In Figure 4e , median RTs (in milliseconds) for correct responses were plotted for each mvASD subgroup and the TD group across all visual stimulus conditions. Strikingly, ‘High’ performers closely mirror the RT pattern of TDs, showing increased reaction time in response to increased stimulus complexity. ‘Low’ performers, however, show consistently elevated RTs across all conditions. Eye-gaze vs. pointing in mvASD ‘low’ performers subgroup To illuminate whether poor pointing performance indicates attenuated perceptual processing we compared pointing and eye-gaze responses of the ‘low’ performance group on a basic b/w saliency oddball condition. We compared ‘transient fixation accuracy’ (TFA) as well as ‘sustained fixation accuracy’ (SFA) with pointing performance with the same criterion (within 200 pixels from oddball target center, see Methods). Analysis revealed a significant difference between TFA and pointing, with participants performing significantly better, when detecting the oddball through eye-gaze (t(8)=3.19, p = 0.013 with a large effect size (d=1.05d) and good discriminatory ability (AUC=0.77, Figure 5a ). We also compared SFA to pointing and found a near-significant trend of more accurate eye gaze (t(8)= 1.99, p= 0.081) with the effect size (d=0.70) indicating a moderate to large difference in performance. Furthermore, the AUC score of 0.69 suggests moderate discriminative ability between eye-gaze vs. pointing ( Figure 5b ). To explore a more liberal criterion, we expanded the radius to 220 pixels. This increased the pointing–TFA gaze difference (t(8)=3.52, p=0.008) and made the SFA comparison significant (t(8)=2.38, p=0.044). However, this radius overlaps distractors, reducing fixation specificity. Thus, the 200-pixel radius was retained as primary. Taken together, superior eye-gaze to pointing performance challenges attenuated visual processing as underlying cause of chance-level performance. Rather, it appears that the visual signal that is perceived does not guide (pointing) behavior. Download figure Open in new tab Figure 5. Eye-gaze vs pointing among ‘low’ performers Each pair of circles connected by a grey line represents one mvASD participant’s eye-gaze and pointing performance respectively for the black/white oddball task. (a) Transient fixation accuracy (TFA) vs. pointing and (b) Sustained fixation accuracy’ (SFA) vs pointing (see methods). Note, the superiority of eye-gaze performance over pointing . Atypical pointing in the Kanizsa oddball task During detection of triangular Kanizsa oddball, off-center pointing was recorded for both children with mvASD and TDs. We compared off-center pointing among successful performers (individual average performance >0.5, chance is ∼0.15). Note that two ‘low’ performers also performed well in this condition ( Figure 2e ). A significant difference was observed between the groups (t(24)= −4.53, p<0.0005). TDs consistently pointed near the center of the Kanizsa triangle, while mvASD participants exhibited a bimodal distribution, pointing either at the center of the triangle or at the center of the Pac-Man inducers ( Figure 6a and 6b ). Consistent with the discrepancy observed in the b/w oddball task between eye-gaze and pointing among “low” performers, the Kanizsa oddball task also reveals that successful perception of the visual target signal appears to drive behavior differently in children with mvASD. Notably, when the Pac-Man inducers were rotated thus removing the illusion, all ‘high’ mvASD performers continued to successfully point to the oddball ( Figure 2e ). However, off-center pointing continued to differ significantly between TDs and mvASD participants (t(23) =-3.89 p = 0.001). Download figure Open in new tab Figure 6. Spontaneous off-center pointing on target Kanizsa oddball (a) Histogram of off-center pointing; (b) Group comparison of off-center pointing. Off-center pointing refers to the pointing distance from the center of the illusory triangle, normalized to the range of 0-1, with ‘0’ = center of the Kanizsa triangle and ‘1’= center of a PacMan shaped inducer. Each circle represents one mvASD or TD participant. Note, in contrast to TDs, mvASD spontaneous pointing appear to be bi-modally distributed around the center of the Kanizsa as well as around the center of the inducer (Pac-Man shape). Standardized descriptive assessments and associations to task performance We examined non-verbal reasoning via the RCPM booklet by pointing and also in puzzle board form. In RCPM booklet 5 out of 18 mvASD participants got a score within age range (∼28%), whereas 9 (50%) scored within age range during puzzle board form version ( Table 1 ). Communication abilities scores assessed via the L-VIS 4.0 ranged from 2-14 with a mean score of 8 (max. Score 17) Seventeen of the 22 participants (77%) were reported to produce unusual sounds (e.g. shrieks) in the L-VIS questionnaire. This included 7 of 11 high performers (64%) and 10 of 11 low performers (91%). Finally, autism diagnosis was confirmed for all with the SCQ ranging from 15 – 30. We analyzed the relationship between visual task performance via pointing and standardized assessment scores ( Figures 7a-7d ). RCPM Puzzle scores were significantly correlated with task accuracy score (R = 0.73 p = 0.001), reflecting a large positive effect size and suggesting that higher cognitive abilities were associated with better performance. Communication ability, LVIS scores, were also positively correlated with accuracy (R = 0.75, p = .0001), indicating a large effect size and that participants with stronger communication skills tended to perform better on the visual tasks. In contrast, perseverative responding on the RCPM Puzzle task was negatively correlated with performance (R = –0.65, p = .005), indicating a large negative effect size and pointing to a possible link between reduced cognitive flexibility and lower task success. No significant correlation was found between SCQ scores and visual performance (R = 0.12, p = 0.61), indicating a negligible effect size and suggesting that social-communication symptom severity was not related to visual task accuracy in this sample. Download figure Open in new tab Figure 7. Correlation between performance and standardized descriptive assessments (a) SCQ, (b) RCPM puzzle, (c) L-VIS and (d) % perseveration in RCPM puzzle. Discussion This study explored early visual processing in minimally verbal autistic children using touchscreen oddball and contour detection tasks measuring pointing and eye-gaze. The methodology eliminated the need for verbal instructions, allowed modelled and guided practice, included short, repeated trials with unlimited response time tailored to the complex behavioral and cognitive profiles of mvASD ( Kasari et al., 2013 ; Matsuzaki et al., 2019 ; Tager-Flusberg et al., 2017 ; Bauminger-Zviely et al., 2020 ). Consistent participation reduced the likelihood that comprehension or task engagement confounded results. Findings revealed notable variability, with distinct low and high performer groups, underscoring that mvASD is not a uniform autism subtype. In general, performance deteriorated with increasing visual complexity, also among some ‘high’ performers. These results could be interpreted as attenuated or atypical early visual processing in mvASD. However, two additional findings may challenge this interpretation: (1) we found that eye-gaze was superior to pointing among “low” performers, with eyes often fixating the oddball item while pointing directed randomly to one of the distractors; (2) The high performers successfully detected the Kanisza oddball, suggesting global perception of this pattern, while at the same time some consistently pointed to the local inducers. It seems that the performance does not solely rely on what they perceive, but also on how they use the visual signal to guide behavior. Heterogeneity and assumed cognitive disability among children with mvASD Children with mvASD have often been grouped into a single subgroup with common skill limitations ( Pizzano et al, 2024 , Kasari et al., 2014 ; Bal et al., 2016 , Ben-Itzchak et al., 2014 ). Several findings challenge assumptions of homogeneity and show evidence of within age range performance for some (e.g. Pizzano et al. 2024 , Courchesne et al, 2015 , Bauminger-Zviely et al., 2020 ). Pointing perfomance in our visual perceptual processing tasks further strengthen heterogeneity within mvASD, ( Figure 4. a, b, c, d and e ). Among others, RTs for ‘high’ performers mirrored TDs’ performance, 50% (n=9) performed within age range on the RCPM puzzle board ( Table 1 ) and notably, lowest performers do well when deviance of target from distractors is larger ( Figure 2. a, d and h ). Furthermore, low performers’ success on some tasks reflects apparent cognitive processing, and that visual stimulus presentation may have jeopardized performance. There were strong positive correlations ( Figure 7a-d ) between visual task performance via pointing with non-verbal fluid reasoning (r = 0.73 p = 0.001, Figure 7b ) and also with communication abilities (r = 0.75, p = .0001, Figure 7c ), suggesting that stronger cognitive and communicative skills support task success. Alternatively, a shared set of underlying, yet unidentified, cognitive mechanisms may be driving the observed strengths across visual task performance, non-verbal reasoning, and communication abilities. In contrast, perseverative responses on the RCPM were negatively correlated with performance (r = –0.65, p = .005, Figure 7d ), indicating that reduced cognitive (or behavioral) flexibility may hinder task accuracy and potentially confound assessments of true intellectual ability. SCQ scores were unrelated to performance (r = 0.12, p = 0.61, Figure 7a ), showing that autism symptom severity did not predict visual task success. Success in our experimental and structured tasks does not necessarily reflect adaptive behavior in fluid, everyday situations (HS clinical observations). Notably, 64% of high performers, even those with high RCPM scores, exhibited unusual vocalizations like shrieks. These vocalizations were a top predictor of variance in adaptive functioning (Naples et al.,2022), highlighting that cognitive strength in structured settings does not necessarily translate to adaptive behavior. Taken together, the current study further supports heterogeneity among children with mvASD and challenges intellectual limitations as the defining factor of these children with common severe verbal limitations. Attenuated visual processing? We show that overall children with mvASD have a distinct behavioral pattern in visual processing tasks, suggesting they rely on low-level features (e.g. orientation). Group performance was better during low-level than mid-level stimuli (e.g. 3D box). Additionally, performance was not exclusively determined by the categorical distinction between low- and mid-level stimuli and recategorizing, by transferring Kanizsa and light from above to low level stimuli improved differentiation (AUC=0.65 to AUC=0.74). Whilst the ability to perceive these illusions require midlevel integration of lower level features, we speculate that some may have succeeded via local low-level features, such as local dark/light opponency between oddball and nearby distractors in light from above, or during the Kanizsa through detection of individual inducers’ smaller angles in the triangular oddball. Saliency effects and paradoxical findings A notable paradox emerged among low performers in the black/white saliency task: they detected a white target more easily among gray distractors than among clearly different black distractors ( Figure 2d ). Since the black and white patches were equally salient relative to the gray background, the children might have selected the target in a saliency map based on local saliency and thus selected a patch at random with the black distractor. In contrast, with gray distracters on a gray background, the white target was more salient and popped out ( Figure 2d ). Perceiving the white target popout among the black distracters, as the high performers did, requires a larger integration window that captures the differences and similarities between patches. Furthermore, the low performers also showed differences in performance during the shape task by better detection of circle than square among triangular distractors, suggesting sensitivity to local shape features (e.g., curves vs. corners). Overall, mvASD visual processing predominantly depends on low-level visual cues, with reduced integration at mid-level complexity. Performance is also influenced by the saliency of stimuli within a visual scene, supporting the notion of saliency maps ( Itti & Koch, 2000 ), where high local contrast and distinctiveness attract attention. These findings align with previous research ( Wang et al., 2015 ), demonstrating stronger attention to low-level features in autism. Dissociation between visual perception and motor behavior At first glance it seems that children with mvASD show atypical early visual processing. However, eye-gaze accuracy outperformed pointing ( Figure 5 ), particularly in transient fixation accuracy (TFA). We also found superior eye gaze to pointing during basic reading skills among young adults with mvASD ( Ellert et al., 2023 ). Furthermore, in the Kanizsa oddball task, mvASD participants who successfully identified the oddball displayed atypical pointing patterns (global triangle center vs. local Pac-Man inducers), unlike TD’s continued pointing to Kanizsa center ( Figure 6 ). Successfully identifying the target Kanizsa oddball can be explained by participants perceiving the global illusory triangular oddball, but pointing was guided by local low-level features. Alternatively, the oddball task was resolved via 2nd order processing and gestalt laws of visual grouping of inducers. This may be a sufficient strategy and is supported by the finding that these differences between mvASD and TDs persisted even when the illusion was disrupted, by rotation of inducers. Previous studies show divergent findings regarding visual perceptual illusions in autism ( Hadad and Yashar, 2022 ). However, we have recently conducted an additional study using the Kanizsa stimuli ( Sykes-Haas and Bonneh, 2025 ) where participants successfully drag correct shape to Kanizsa and shift eye gaze and point to center, with minor visual modulations, suggesting that the illusion is perceived even though behavior suggests otherwise. Finally, the increase in number of participants reaching age level performance during RCPM puzzle as opposed to pointing in booklet further emphasizes assessment by pointing may not uncover true perception (or cognition). Taken together, these findings, of better eye gaze among ‘low’ performers and atypical Kanizsa pointing among ‘high’ performers challenge notions of attenuated visual processing, showing a dissociation between visual perception and motor behavior and highlighting differences in how visual signals guide behavior in mvASD. Suggested accounts for the perception-action dissociation in mvASD What could account for this hand-eye discrepancy? We briefly outline several theories and findings that might offer plausible explanations. Whilst far from exhaustive, this brief overview aims to provide directions for future investigation and hypothesis generation. One possible explanation for the poor pointing performance in mvASD is increased motor variability. Torres et al. (2013) reported elevated stochastic variability in both goal-directed and goal-less hand movements in ASD, particularly among nonverbal individuals, indicating reduced predictability and increased motor noise. We recently extended this to oculomotor behavior, showing increased saccadic variability correlated with ADOS scores ( Ziv et al., 2024 ). This instability, reflected in frequent short inter-saccade intervals, may lead to impulsive gaze shifts. As saccades guide hand movements (Neggers et al., 2000), such variability may cause mvASD children to miss targets their initial transient fixations (TFA) briefly located. Donk et al. (2008) found that in neurotypical adults, fast saccades reliably target the most salient stimulus, while delayed ones show more errors, they suggest, that “saliency is short lived” and fades within a few hundred milliseconds. Thus, delayed responses shift from stimulus-to goal-driven , increasing errors. Pupil dilation, an indicator of LC–NE system activity, reflects arousal that supports goal-directed behavior ( London, 2018 ). Reduced pupil dilation in mvASD ( Ellert et al., 2023 ; Bonneh et al., 2025 ) may suggest insufficient arousal recruitment and explain why accurate initial fixation ( Figure 5a ) is followed by less precise goal directed pointing. Lidstone and Mostofsky (2021) suggest autism involves disrupted visual-motor integration (VMI), with greater reliance on proprioception and visual feedback often hindering goal-directed actions. While our findings support a VMI dissociation, they refine this model by showing that visual input can aid behavior— particularly when targets are more salient, deviant, or defined by low-level features. Despite identical motor demands in all tasks (pointing to a target), performance varied with visual characteristics, and a gaze–pointing dissociation showed targets were seen but not selected. Similarly, Torres et al. (2013) found that stimulus type affects stochastic variability in pointing. These results suggest visual input isn’t inherently disruptive but modulates behavior based on its properties, extending Mostofsky’s model to account for variability in how visual information influences VMI. Zooming out Cisek’s affordance competition hypothesis (ACH; Cisek, 2007 ) suggests the dorsal stream encodes multiple potential actions/affordances (e.g., six in our oddball tasks) that compete for selection based on task relevance. This model may clarify perception–behavior dissociation in mvASD. Disruptions in affordance competition may underlie atypical sensory–motor integration and inconsistent goal-directed behavior. Contributing factors may include aforementioned inefficient arousal, motor variability, stimulus-driven attention dominance, and visual–motor integration deficits. Together, these may dysregulate affordance competition and lead to inconsistent motor output in mvASD (see also Robertson and Baron-Cohen, 2017 on ambiguity resolution). Thus, whilst our findings suggest that children with mvASD may “see the point,” they do not always point to what they see. Clinical implications, limitations and future studies? Despite small sample-size, consistent response patterns indicate that visual stimulus type affects mvASD performance. Visual aids are widely used in mvASD intervention (e.g., Arthur-Kelly et al., 2009 ), and as the AG case in the introduction illustrates, the way visual information is presented can shape (communicative) success. We explored how processing level, deviance, and saliency modulate pointing, highlighting the need for well-designed visual materials. Future studies could develop a standardized set of visual-perceptual stimuli to assess, prior to intervention or experiment, how visual features affect behavior in mvASD. Our observation that eye gaze responses were more robust than pointing also carries clinical implications and should be explored further. We also speculate that ‘perceptual load’ from complex (e.g., 3D box) or ambiguous (e.g., Kanizsa) stimuli affects motor output in mvASD. Consistent with this, Torres et al. (2013) showed reduced micro-movements with geometric shapes in nonverbal autistic children, and Thorpe et al. (2021) linked cognitive load to increased corrective movements in neurotypicals. We propose a systematic investigation into visual stimulus levels’ (i.e. ‘perceptual load’) modulation of goal-directed pointing performance. This could clarify perception–action links and support personalized interventions. Summary and conclusions This study examined early visual processing in children with mvASD using nonverbal touchscreen tasks. Results showed marked variability in performance, challenging the assumption of uniform impairments. Whilst performance was generally stronger for low-level visual stimuli, it was also influenced by saliency and contrast even within low level stimuli, underscoring the role of the saliency map and the visual presentation layout. A key finding was the dissociation between perception and motor output: e.g. participants fixated correctly on targets but failed to point to them. This suggests intact visual perception but disruptions in translating perception into action. We interpret this through the lens of the affordance competition hypothesis ( Cisek, 2007 )—mvASD individuals may perceive multiple potential actions but struggle to select and execute the appropriate one, possibly due to motor randomness, attentional impulsivity, impaired transient arousal, over-reliance on local visual saliency, and/or inefficient visual–motor integration. These findings highlight the value of eye gaze measures and call for research into how perceptual input affects motor demands. ETHICS STATEMENT The study was approved by the Chief Scientist at the Israeli Ministry of Education and the Bar Ilan University internal review board committee. Written informed consent was obtained from all parents. All participants’ legal guardians provided written informed consent. DATA AVAILABILITY STATEMENT The data and materials will be available upon reasonable request from the corresponding author Yoram S. Bonneh. FUNDING This research was supported by the Israel Science Foundation (Grant No. 657/21). CONFLICT OF INTEREST The authors declare no conflicts of interest. ACKNOWLEDGMENTS We thank the participating children and their families for their invaluable contribution to this study. Appendix 1 Individual participants’ descriptive measures View this table: View inline View popup Download powerpoint Funder Information Declared Israel Science Foundation, https://ror.org/04sazxf24 , Grant No. 657/21 References American Psychiatric Association . ( 2013 ). Diagnostic and statistical manual of mental disorders (5th ed .) ↵ Arthur-Kelly M , Sigafoos J , Green V , Mathisen B , Arthur-Kelly R. Issues in the use of visual supports to promote communication in individuals with autism spectrum disorder . Disabil Rehabil . 2009 ; 31 ( 18 ): 1474 – 86 . doi: 10.1080/09638280802590629 . PMID: 19296313 . OpenUrl CrossRef PubMed ↵ Bal VH , Katz T , Bishop SL , Krasileva K. Understanding definitions of minimally verbal across instruments: evidence for subgroups within minimally verbal children and adolescents with autism spectrum disorder . J Child Psychol Psychiatry . 2016 Dec ; 57 ( 12 ): 1424 – 1433 . doi: 10.1111/jcpp.12609 . Epub 2016 Jul 30. PMID: 27473432 . OpenUrl CrossRef PubMed ↵ Bauminger-Zviely N , Estrugo Y , Samuel-Magal K , Friedlin A , Heishrik L , Koren D , Bar-Yehuda S. Communicating Without Words: School-Based RCT Social Intervention in Minimally Verbal Peer Dyads with ASD . J Clin Child Adolesc Psychol . 2020 Nov-Dec ; 49 ( 6 ): 837 – 853 . doi: 10.1080/15374416.2019.1660985 . Epub 2019 Sep 27. PMID: 31560567 . OpenUrl CrossRef PubMed ↵ Ben-Itzchak E , Watson LR , Zachor DA . Cognitive ability is associated with different outcome trajectories in autism spectrum disorders . J Autism Dev Disord . 2014 Sep ; 44 ( 9 ): 2221 – 9 . doi: 10.1007/s10803-014-2091-0 . PMID: 24710810 . OpenUrl CrossRef PubMed Behrmann M , Thomas C , Humphreys K. Seeing it differently: visual processing in autism . Trends Cogn Sci . 2006 Jun ; 10 ( 6 ): 258 – 64 . doi: 10.1016/j.tics.2006.05.001 . Epub 2006 May 19. PMID: 16713326 . OpenUrl CrossRef PubMed Web of Science ↵ Bertone A , Mottron L , Jelenic P , Faubert J. Motion perception in autism: a “complex” issue . J Cogn Neurosci . 2003 Feb 15; 15 ( 2 ): 218 – 25 . doi: 10.1162/089892903321208150 . PMID: 12676059 . OpenUrl CrossRef PubMed Web of Science ↵ Bertone A , Mottron L , Jelenic P , Faubert J. Enhanced and diminished visuo-spatial information processing in autism depends on stimulus complexity . Brain . 2005 Oct ; 128 ( Pt 10 ): 2430 - 41 . doi: 10.1093/brain/awh561 . Epub 2005 Jun 15. PMID: 15958508 . OpenUrl CrossRef PubMed Web of Science ↵ Bonneh Y , Sagi D. ( 1998 ) Effects of spatial configuration on contrast detection . Vision Res .; 38 ( 22 ): 3541 – 53 . doi: 10.1016/s0042-6989(98)00045-5 . PMID: 9893788 . OpenUrl CrossRef PubMed Web of Science ↵ Bonneh , Y.S , Sykes-Haas , H.S , Ellert , K. ( 2025 ) Seeing but Not Always Pointing: Hidden Perceptual and Cognitive Skills in Minimally Verbal Autism . THE 11th Conference on Cognition Research (ISCOP), Akko, Israel (Abstract) Booth RDL , Happé FGE. Evidence of Reduced Global Processing in Autism Spectrum Disorder . J Autism Dev Disord . 2018 Apr ; 48 ( 4 ): 1397 – 1408 . doi: 10.1007/s10803-016-2724-6 . PMID: 26864159 ; PMCID: PMC5861162 . OpenUrl CrossRef PubMed ↵ Brignell A , Chenausky KV , Song H , Zhu J , Suo C , Morgan AT . Communication interventions for autism spectrum disorder in minimally verbal children . Cochrane Database Syst Rev . 2018 Nov 5; 11 ( 11 ): CD012324 . doi: 10.1002/14651858.CD012324.pub2 . PMID: 30395694 ; PMCID: PMC6516977 . OpenUrl CrossRef PubMed ↵ Cisek P. Cortical mechanisms of action selection: the affordance competition hypothesis . Philos Trans R Soc Lond B Biol Sci . 2007 Sep 29; 362 ( 1485 ): 1585 – 99 . doi: 10.1098/rstb.2007.2054 . PMID: 17428779 ; PMCID: PMC2440773 . OpenUrl CrossRef PubMed ↵ Courchesne V , Meilleur AA , Poulin-Lord MP , Dawson M , Soulières I. Autistic children at risk of being underestimated: school-based pilot study of a strength-informed assessment . Mol Autism . 2015 Mar 6; 6 : 12 . doi: 10.1186/s13229-015-0006-3 . PMID: 25774281 ; PMCID: PMC4359559 . OpenUrl CrossRef PubMed ↵ Dakin S , Frith U. Vagaries of visual perception in autism . Neuron . 2005 Nov 3; 48 ( 3 ): 497 – 507 . doi: 10.1016/j.neuron.2005.10.018 . PMID: 16269366 . OpenUrl CrossRef PubMed Web of Science ↵ Dawson M , Soulières I , Gernsbacher MA , Mottron L. The level and nature of autistic intelligence . Psychol Sci . 2007 Aug ; 18 ( 8 ): 657 – 62 . doi: 10.1111/j.1467-9280.2007.01954.x . PMID: 17680932 ; PMCID: PMC4287210 . OpenUrl CrossRef PubMed Web of Science Donk M , van Zoest W. Effects of salience are short-lived . Psychol Sci . 2008 Jul ; 19 ( 7 ): 733 – 9 . doi: 10.1111/j.1467-9280.2008.02149.x . PMID: 18727790 . OpenUrl CrossRef PubMed Web of Science ↵ Ellert , K , Shefer-Kaufman , N and Bonneh , Y.S. ( 2023 ) Can Minimally Verbal Individuals with Autism Read? Hidden Literacy Revealed Via Eye Movements , International Society for Autism Research-Annual Meeting (INSAR), Stockholm, Sweden (abstract) . ↵ Field DJ , Hayes A , Hess RF . Contour integration by the human visual system: evidence for a local “association field” . Vision Res . 1993 Jan ; 33 ( 2 ): 173 – 93 . doi: 10.1016/0042-6989(93)90156-q . PMID: 8447091 . OpenUrl CrossRef PubMed Web of Science Frith U , Happé F. Autism: beyond “theory of mind” . Cognition . 1994 Apr-Jun ; 50 ( 1-3 ): 115 – 32 . doi: 10.1016/0010-0277(94)90024-8 . PMID: 8039356 . OpenUrl CrossRef PubMed Web of Science ↵ Girard D , Courchesne V , Cimon-Paquet C , Jacques C , Soulières I. Visual abilities and exploration behaviors as predictors of intelligence in autistic children from preschool to school age . Autism . 2023 Nov ; 27 ( 8 ): 2446 – 2464 . doi: 10.1177/13623613231166189 . Epub 2023 Apr 27. PMID: 37113016 . OpenUrl CrossRef PubMed Gliga T , Bedford R , Charman T , Johnson MH ; BASIS Team. Enhanced Visual Search in Infancy Predicts Emerging Autism Symptoms . Curr Biol . 2015 Jun 29; 25 ( 13 ): 1727 – 30 . doi: 10.1016/j.cub.2015.05.011 . Epub 2015 Jun 11. PMID: 26073135 ; PMCID: PMC4502951 . OpenUrl CrossRef PubMed ↵ Haaf RA , Fulkerson AL , Jablonski BJ , Hupp JM , Shull SS , Pescara-Kovach L. Object recognition and attention to object components by preschool children and 4-month-old infants . J Exp Child Psychol . 2003 Oct ; 86 ( 2 ): 108 – 23 . doi: 10.1016/s0022-0965(03)00108-5 . PMID: 13129698 . OpenUrl CrossRef PubMed ↵ Hadad BS , Yashar A. Sensory Perception in Autism: What Can We Learn? Annu Rev Vis Sci . 2022 Sep 15; 8 : 239 – 264 . doi: 10.1146/annurev-vision-093020-035217 . Epub 2022 Jul 8. PMID: 35804481 . OpenUrl CrossRef PubMed ↵ Hayes , G. R. , Hirano , S. , Marcu , G. , Monibi , M. , Nguyen , D. H. , & Yeganyan , M. ( 2010 ). Interactive visual supports for children with autism . Personal and ubiquitous computing , 14 , 663 – 680 . OpenUrl CrossRef ↵ Itti L , Koch C. A saliency-based search mechanism for overt and covert shifts of visual attention . Vision Res . 2000 ; 40 ( 10-12 ): 1489 – 506 . doi: 10.1016/s0042-6989(99)00163-7 . PMID: 10788654 . OpenUrl CrossRef PubMed Web of Science ↵ Kasari C , Brady N , Lord C , Tager-Flusberg H. Assessing the minimally verbal school-aged child with autism spectrum disorder . Autism Res . 2013 Dec ; 6 ( 6 ): 479 – 93 . doi: 10.1002/aur.1334 . Epub 2013 Oct 29. PMID: 24353165 ; PMCID: PMC4139180 . OpenUrl CrossRef PubMed Web of Science ↵ Kasari , C. , Kaiser , A. , Goods , K. , Nietfeld , J. , Mathy , P. , Landa , R. , … & Almirall , D. ( 2014 ). Communication interventions for minimally verbal children with autism: A sequential multiple assignment randomized trial . Journal of the American Academy of Child & Adolescent Psychiatry , 53 ( 6 ), 635 – 646 . OpenUrl CrossRef PubMed Kéïta L , Guy J , Berthiaume C , Mottron L , Bertone A. An early origin for detailed perception in Autism Spectrum Disorder: biased sensitivity for high-spatial frequency information . Sci Rep . 2014 Jul 4; 4 : 5475 . doi: 10.1038/srep05475 . PMID: 24993026 ; PMCID: PMC4081897 . OpenUrl CrossRef PubMed ↵ Lidstone DE , Mostofsky SH . Moving Toward Understanding Autism: Visual-Motor Integration, Imitation, and Social Skill Development . Pediatr Neurol . 2021 Sep ; 122 : 98 – 105 . doi: 10.1016/j.pediatrneurol.2021.06.010 . Epub 2021 Jul 28. PMID: 34330613 ; PMCID: PMC8372541 . OpenUrl CrossRef PubMed ↵ London EB . Neuromodulation and a Reconceptualization of Autism Spectrum Disorders: Using the Locus Coeruleus Functioning as an Exemplar . Front Neurol . 2018 Dec 19; 9 : 1120 . doi: 10.3389/fneur.2018.01120 . PMID: 30619071 ; PMCID: PMC6305710 . OpenUrl CrossRef PubMed Low Verbal Investigatory Survey (LVIS) Project Home : ( https://autismlanguagelab.psychology.uconn.edu/low-verbal-investigatory-survey-elvis-project-home/ ) ↵ Manassi M , Sayim B , Herzog MH . When crowding of crowding leads to uncrowding . J Vis . 2013 Nov 8; 13 ( 13 ): 10 . doi: 10.1167/13.13.10 . PMID: 24213598 . OpenUrl Abstract / FREE Full Text ↵ Matsuzaki J , Kuschner ES , Blaskey L , Bloy L , Kim M , Ku M , Edgar JC , Embick D , Roberts TPL . Abnormal auditory mismatch fields are associated with communication impairment in both verbal and minimally verbal/nonverbal children who have autism spectrum disorder . Autism Res . 2019 Aug ; 12 ( 8 ): 1225 – 1235 . doi: 10.1002/aur.2136 . Epub 2019 May 28. PMID: 31136103 ; PMCID: PMC6710834 . OpenUrl CrossRef PubMed ↵ McKyton A , Ben-Zion I , Doron R , Zohary E. The Limits of Shape Recognition following Late Emergence from Blindness . Curr Biol . 2015 Sep 21; 25 ( 18 ): 2373 – 8 . doi: 10.1016/j.cub.2015.06.040 . Epub 2015 Aug 20. PMID: 26299519 . OpenUrl CrossRef PubMed ↵ Mottron L , Dawson M , Soulières I , Hubert B , Burack J. Enhanced perceptual functioning in autism: an update, and eight principles of autistic perception . J Autism Dev Disord . 2006 Jan ; 36 ( 1 ): 27 – 43 . doi: 10.1007/s10803-005-0040-7 . PMID: 16453071 .. OpenUrl CrossRef PubMed Web of Science Naples A , Tenenbaum EJ , Jones RN , Righi G , Sheinkopf SJ , Eigsti IM . Exploring communicative competence in autistic children who are minimally verbal: The Low Verbal Investigatory Survey for Autism (LVIS) . Autism . 2023 Jul ; 27 ( 5 ): 1391 – 1406 . doi: 10.1177/13623613221136657 . Epub 2022 Nov 14. PMID: 36373838 ; PMCID: PMC10183057 . OpenUrl CrossRef PubMed Neggers SF , Bekkering H. Ocular gaze is anchored to the target of an ongoing pointing movement . J Neurophysiol . 2000 Feb ; 83 ( 2 ): 639 – 51 . doi: 10.1152/jn.2000.83.2.639 . PMID: 10669480 . OpenUrl CrossRef PubMed Web of Science ↵ Nishimura M , Scherf S , Behrmann M. Development of object recognition in humans . F1000 Biol Rep . 2009 Jul 27; 1 : 56 . doi: 10.3410/B1-56 . PMID: 20948628 ; PMCID: PMC2948260 . OpenUrl CrossRef PubMed ↵ Ostrolenk A , Forgeot d’Arc B , Jelenic P , Samson F , Mottron L. Hyperlexia: Systematic review, neurocognitive modelling, and outcome . Neurosci Biobehav Rev . 2017 Aug ; 79 : 134 – 149 . doi: 10.1016/j.neubiorev.2017.04.029 . Epub 2017 May 3. PMID: 28478182 . OpenUrl CrossRef PubMed ↵ Pizzano M , Shire S , Shih W , Levato L , Landa R , Lord C , Smith T , Kasari C. Profiles of minimally verbal autistic children: Illuminating the neglected end of the spectrum . Autism Res . 2024 Jun ; 17 ( 6 ): 1218 – 1229 . doi: 10.1002/aur.3151 . Epub 2024 May 27. PMID: 38803132 ; PMCID: PMC11186722 . OpenUrl CrossRef PubMed ↵ Plesa Skwerer D , Jordan SE , Brukilacchio BH , Tager-Flusberg H. Comparing methods for assessing receptive language skills in minimally verbal children and adolescents with autism spectrum disorders . Autism . 2016 Jul ; 20 ( 5 ): 591 – 604 . doi: 10.1177/1362361315600146 . Epub 2015 Sep 25. PMID: 26408635 ; PMCID: PMC6988895 . OpenUrl CrossRef PubMed Posar A , Visconti P. Update about “minimally verbal” children with autism spectrum disorder . Rev Paul Pediatr . 2021 Sep 1; 40 : e2020158 . doi: 10.1590/1984-0462/2022/40/2020158 . PMID: 34495269 ; PMCID: PMC8432069 . OpenUrl CrossRef PubMed ↵ Raven , J. C. , Raven , J. , & Court , J. ( 1998 ). Manual for Raven’s Coloured Progressive Matrices. London, England: H. K. Western Psychological Services ↵ Rose V , Trembath D , Keen D , Paynter J. The proportion of minimally verbal children with autism spectrum disorder in a community-based early intervention programme . J Intellect Disabil Res . 2016 May ; 60 ( 5 ): 464 – 77 . doi: 10.1111/jir.12284 . PMID: 27120989 . OpenUrl CrossRef PubMed ↵ Rutter , M. , Bailey , A. , & Lord , C. ( 2003 ). The Social Communication Questionnaire manual . Los Angeles, CA : Western Psychological Services . ↵ Simmons DR , Robertson AE , McKay LS , Toal E , McAleer P , Pollick FE . Vision in autism spectrum disorders . Vision Res . 2009 Nov ; 49 ( 22 ): 2705 – 39 . doi: 10.1016/j.visres.2009.08.005 . Epub 2009 Aug 12. PMID: 19682485 . OpenUrl CrossRef PubMed Web of Science ↵ Sykes-Haas , H.S. and Bonneh , Y.S. ( 2025 ) Minimally Verbal Children with Autism ‘See the Point, but do not Point to What They See’: A Behavioral and Eye-Tracking in Visual Perceptual Processing . International Society for Autism Research-Annual Meeting (INSAR), Seattle, Washinton. (abstract) ↵ Tager-Flusberg H , Plesa Skwerer D , Joseph RM , Brukilacchio B , Decker J , Eggleston B , Meyer S , Yoder A. Conducting research with minimally verbal participants with autism spectrum disorder . Autism . 2017 Oct ; 21 ( 7 ): 852 – 861 . doi: 10.1177/1362361316654605 . Epub 2016 Jun 26. PMID: 27354431 ; PMCID: PMC6988898 . OpenUrl CrossRef PubMed ↵ Tager-Flusberg H , Kasari C. Minimally verbal school-aged children with autism spectrum disorder: the neglected end of the spectrum . Autism Res . 2013 Dec ; 6 ( 6 ): 468 – 78 . doi: 10.1002/aur.1329 . Epub 2013 Oct 7. PMID: 24124067 ; PMCID: PMC3869868 . OpenUrl CrossRef PubMed Web of Science Thorpe , A. , Friedman , J. , Evans , S. , Nesbitt , K. , & Eidels , A. ( 2022 ). Mouse movement trajectories as an indicator of cognitive workload . International Journal of Human–Computer Interaction , 38 ( 15 ), 1464 – 1479 . OpenUrl CrossRef ↵ Torres EB , Brincker M , Isenhower RW , Yanovich P , Stigler KA , Nurnberger JI , Metaxas DN , José JV. Autism: the micro-movement perspective . Front Integr Neurosci . 2013 Jul 24; 7 : 32 . doi: 10.3389/fnint.2013.00032 . PMID: 23898241 ; PMCID: PMC3721360 . OpenUrl CrossRef PubMed ↵ Wang S , Jiang M , Duchesne XM , Laugeson EA , Kennedy DP , Adolphs R , Zhao Q. Atypical Visual Saliency in Autism Spectrum Disorder Quantified through Model-Based Eye Tracking . Neuron . 2015 Nov 4; 88 ( 3 ): 604 – 16 . doi: 10.1016/j.neuron.2015.09.042 . Epub 2015 Oct 22. PMID: 26593094 ; PMCID: PMC4662072 . OpenUrl CrossRef PubMed ↵ Wass SV , Jones EJ , Gliga T , Smith TJ , Charman T , Johnson MH ; BASIS team. Shorter spontaneous fixation durations in infants with later emerging autism . Sci Rep . 2015 Feb 6; 5 : 8284 . doi: 10.1038/srep08284 . PMID: 25655672 ; PMCID: PMC4319149 . OpenUrl CrossRef PubMed ↵ Ziv I , Avni I , Dinstein I , Meiri G , Bonneh YS . Oculomotor randomness is higher in autistic children and increases with the severity of symptoms . Autism Res . 2024 Feb ; 17 ( 2 ): 249 – 265 . doi: 10.1002/aur.3083 . Epub 2024 Jan 8. PMID: 38189581 . OpenUrl CrossRef PubMed View the discussion thread. Back to top Previous Next Posted June 28, 2025. Download PDF Supplementary Material Email Thank you for your interest in spreading the word about bioRxiv. NOTE: Your email address is requested solely to identify you as the sender of this article. Your Email * Your Name * Send To * Enter multiple addresses on separate lines or separate them with commas. You are going to email the following Minimally verbal children with autism may ‘see the point, but do not (always) point to what they see’ Message Subject (Your Name) has forwarded a page to you from bioRxiv Message Body (Your Name) thought you would like to see this page from the bioRxiv website. Your Personal Message CAPTCHA This question is for testing whether or not you are a human visitor and to prevent automated spam submissions. Share Minimally verbal children with autism may ‘see the point, but do not (always) point to what they see’ Hannah S. Sykes-Haas , Yoram S. Bonneh bioRxiv 2025.06.26.661808; doi: https://doi.org/10.1101/2025.06.26.661808 Share This Article: Copy Citation Tools Minimally verbal children with autism may ‘see the point, but do not (always) point to what they see’ Hannah S. Sykes-Haas , Yoram S. Bonneh bioRxiv 2025.06.26.661808; doi: https://doi.org/10.1101/2025.06.26.661808 Citation Manager Formats BibTeX Bookends EasyBib EndNote (tagged) EndNote 8 (xml) Medlars Mendeley Papers RefWorks Tagged Ref Manager RIS Zotero Tweet Widget Facebook Like Google Plus One Subject Area Neuroscience Subject Areas All Articles Animal Behavior and Cognition (7640) Biochemistry (17707) Bioengineering (13903) Bioinformatics (41980) Biophysics (21465) Cancer Biology (18613) Cell Biology (25528) Clinical Trials (138) Developmental Biology (13387) Ecology (19920) Epidemiology (2067) Evolutionary Biology (24332) Genetics (15615) Genomics (22519) Immunology (17747) Microbiology (40424) Molecular Biology (17194) Neuroscience (88664) Paleontology (667) Pathology (2839) Pharmacology and Toxicology (4827) Physiology (7650) Plant Biology (15160) Scientific Communication and Education (2046) Synthetic Biology (4302) Systems Biology (9826) Zoology (2271)