Skip to main content
Neurology logoLink to Neurology
. 2023 Feb 7;100(6):e639–e650. doi: 10.1212/WNL.0000000000201496

Relationship of Impairments in Associative Learning With Intellectual Disability and Cerebellar Hypoplasia in Autistic Children

John P Welsh 1,, Jeffrey Munson 1, Tanya St John 1, Christina N Meehan 1, Elise Tran Abraham 1, Frederick B Reitz 1, K Kawena Begay 1, Stephen R Dager 1, Annette M Estes 1
PMCID: PMC9946191  PMID: 36443015

Abstract

Background and Objectives

The severity of autism spectrum disorder (ASD) varies widely and is associated with intellectual disability (ID) and brain dysmorphology. We tested the hypothesis that the heterogeneity of ASD can be accounted for, in part, by altered associative learning measured by eye-blink conditioning (EBC) paradigms, used to test for forebrain and cerebellar dysfunction across the full range of ASD severity and intellectual ability.

Methods

Children in this cohort study were diagnosed with ASD or typical development (TD); most children were recruited from a 10-year longitudinal study. Outcome measures were the percentage and timing of conditioned eye-blink responses (CRs) acquired to a tone, recorded photometrically and related to measures of ASD severity, IQ, and age 2 brain morphometry by MRI. A sequence of trace and delay EBC was used. Analysis of variance, t test, and logistic regression (LR) were used.

Results

Sixty-two children were studied at school age. Nine children with ASD with ID since age 2 (ASD + ID; IQ = 49 ± 6; 11.9 ± 0.2 years old [±SD]) learned more slowly than 30 children with TD (IQ = 120 ± 16; 10.5 ± 1.5 years old [±SD]) during trace EBC and showed atypically early-onset CRs (1.4 SD pre-TD) related to hypoplasia of the cerebellum at age 2 but not of the amygdala, hippocampus, or cerebral cortex. Conversely, 16 children with ASD with robust intellectual development since age 2 (IQ = 100 ± 3; 12.0 ± 0.4 years old [±SD]) learned typically but showed early-onset CRs only during long-delay EBC (0.8 SD pre-TD) unrelated to hypoplasia of any measured brain area. Using 16 EBC measures, binary LR classified ASD and TD with 80% accuracy (95% CI = 72–88%), 81% sensitivity (95% CI = 69–92%), and 79% specificity (95% CI = 68–91%); multinomial LR more accurately classified children based on ID (94% accuracy, 95% CI = 89–100%) than ASD severity (85% accuracy, 95% CI = 77–93%). Separate analyses of 39 children with MRI (2.1 ± 0.3 years old [±SD]) indicated that cerebellar hypoplasia did not predict ASD + ID over ages 2–4 (Cohen d = 0.3) compared with early-onset CRs during age 11 trace EBC (Cohen d = −1.3).

Discussion

Trace EBC reveals the relationship between cerebellar hypoplasia and ASD + ID likely by engaging cerebrocerebellar circuits involved in intellectual ability and implicit timing. Follow-up prospective studies using associative learning can determine whether ID can be predicted in children with early ASD diagnoses.


Autism spectrum disorder (ASD) is a neurodevelopmental disability with core features of impaired social communication plus restricted interests and repetitive behaviors.1 The severity of ASD varies extensively and can include significant intellectual disability (ID), indicating marked heterogeneity in the brain mechanisms underlying ASD. Changes in brain morphology are promising early childhood markers of ASD.2 Identifying individualized changes in brain functioning by behavioral measures that can be linked to the functioning of neuroanatomic structures may help early detection and suggest individualized interventions to modify the progression of ASD-related disability.

Cerebellar hypoplasia was an early documented change in brain morphology in ASD,3,4 although its meaning remains elusive. Alterations in cerebellar development may contribute to ASD feature heterogeneity, extrapolating from contemporary studies demonstrating the involvement of posterior lobe cerebellum in human cognition.5-8 Cerebellar hypoplasia with ASD has not been uniformly replicated,9-12 perhaps reflecting participant heterogeneity across studies.13 Eye-blink conditioning (EBC) is a family of associative learning paradigms that engage partly nonoverlapping forebrain structures14-17 in which the subsecond timing and amplitude of conditioned responses (CRs) is impaired by cerebellar dysfunction.18-21 Three studies have indicated that EBC may help elucidate the involvement of the cerebellum in ASD.22-24

The objective of our study was therefore to examine the relationships between EBC measures of learning, the severity of ASD clinical features, and early-childhood morphology of brain structures implicated in learning. We hypothesized that heterogeneity in ASD severity, ID, and early childhood brain morphology could be resolved by heterogeneity in associative learning using EBC paradigms as neurologic assessments.

Methods

Standard Protocol Approvals, Registrations, and Patient Consents

The studies were approved by the University of Washington Human Subjects Division Institutional Review Board and performed according to the Declaration of Helsinki. Participants' guardians gave written informed consent, and participants gave verbal assent when possible. Participants were free to withdraw at any time.

Participants and ASD Diagnoses

A total of 83 children with ASD or typical development (TD) were studied with racial demographics of Seattle, United States (70% White, 11% Asian, 11% Black or Multiracial, and 8% Latino). Participants were 76% male, consistent with the ASD sex distribution. ASD diagnoses were made by licensed clinical psychologists at the University of Washington Autism Center using contemporaneous diagnostic criteria (Diagnostic Statistical Manual [DSM]-IV, 2004–2010; DSM-V, 2014–2017) and were informed by the Autism Diagnostic Observation Schedule (ADOS). Exclusion criteria are detailed in eMethods (links.lww.com/WNL/C490). Children participated in one or both of 2 studies (Figure 1A).

Figure 1. Experimental Flow Diagram and EBC Paradigms.

Figure 1

(A) Flow diagram showing participants and sample sizes for the experiment. Study 1 investigated a cohort of children with ASD and TD (n = 31) that was recruited in 2004–2007 and assessed longitudinally over ages 2–12 years (Figure 2). Comparisons of EBC measures from the longitudinal ASD cohort at age 11–12 were made to a larger age- and sex-matched TD cohort (n = 30) (Figure 3). Cross-sectional analyses were performed using binary LR (Figure 4, A and B) by combining the longitudinal ASD cohort (n = 25) with children who were ascertained as having ASD and assessed identically at age 11–12 (n = 7). Multinomial LR was performed using the longitudinal ASD cohort and the TD control group when ID was the dependent variable (Figure 4C, left) or using the full cross-sectional cohort when ASD severity was the dependent variable (Figure 4C, right). Study 2 consisted of 39 children who received brain structure MRI at age 2 in 2004–2007 (n = 27 ASD; n = 12 TD); some with ASD participated in study 1 at age 11 years (n = 15) while the balance (n = 12) was lost to follow-up after age 4. All children with ASD in study 2 were studied annually over ages 2–4 years with diagnostic assessments. Brain morphology from the subset that progressed to EBC at age 11 is presented in Figures 5 and 6. Brain morphology and diagnostic assessments for the entire study 2 ASD cohort are presented in Figure 7, A and B and eFigure 3 (links.lww.com/WNL/C490). (B–D) Mean CR waveforms ±SEM (dotted lines) of 30 participants with TD for 3 EBC paradigms. Block diagrams show the stimuli and stimulus timing for trace (B), long-delay (C), and short-delay (D) EBC. Arrows indicate CR measures. Upward deflection represents eyelid closure. The horizontal dotted lines indicate the CR onset threshold (baseline mean +1 SD of baseline). ASD = autism spectrum disorder; ID = intellectual disability; EBC = eye-blink conditioning; TD = typical development.

Study 1: EBC Across the Full ASD Spectrum

Study 1 involved 62 children who were assessed clinically and studied with EBC at age 11.2 ± 0.2 years (n = 32 ASD, 81% male, n = 30 TD, 71% male). Of that group, 31 children (n = 25 ASD, n = 6 TD) had previously participated in 2 longitudinal studies at ages 2, 3, 4, and 6 years.25,26 The remaining children (n = 7 ASD; n = 24 TD) were ascertained for cross-sectional analyses while matching for sex across experimental groups. Intellectual ability was assessed by the Mullen Scales of Early Learning (MSEL) (ages 2–4),27 the Differential Abilities Scales (DAS) early years (age 6), and DAS-II (age 10–12).28 ASD severity was assessed by the ADOS-Western Psychological Services (ADOS) (ages 2–6) and the ADOS-2 (age 10–12).29 Adaptive behavior was assessed using the Vineland Adaptive Behavior Scale, 2nd Edition (age 10–12).30 Information regarding prescription drug use, affiliated medical conditions, and gestational age at birth is provided in eMethods (links.lww.com/WNL/C490).

EBC consists of paradigms in which a child associates a tone conditioned stimulus (CS) with a corneal air-puff unconditioned stimulus (US).31 Before conditioning, the US elicits a reflex eye blink that is termed the unconditioned response. Across trials in which the CS precedes the US at a fixed CS-US interval, children acquire an eye-blink to the CS termed the conditioned response (CR). CRs acquired to the CS exhibit onsets and peak amplitudes with latencies that adapt to the CS-US interval, thereby covering the eye at US onset.

Figure 1B–D shows the conditioning paradigms and measures. Trace and delay EBC occurred over 4 sessions (90 trials/session; 20 ± 5 s intertrial interval; 14 ± 2 d between sessions).23,24 Trace EBC is defined by a stimulus-free period during the CS-US interval. Delay EBC is defined by a CS that extends through the CS-US interval. Sessions 1 and 2 used trace EBC at a 700-ms CS-US interval (200-ms CS followed by a 500-ms silent period and the US; Figure 1B). Session 3 used long-delay EBC (700 ms CS-US interval, 800 ms CS, coterminating US; Figure 1C). Session 4 used short-delay EBC (400 ms CS-US interval, 500 ms CS, coterminating US; Figure 1D). We did not perform short-delay EBC for 1 participant with ASD. Long-delay EBC data for 1 participant with TD were not collected due to a technical issue. Full EBC methodology is provided in eMethods (links.lww.com/WNL/C490).

Study 2: Age 2 Brain Morphometry

Study 2 involved 39 children who underwent brain structural MRI at age 2.1 ± 0.04 years (n = 27 ASD, 78% male; n = 12 TD, 83% male). The MRI data were included in a study of T2 quantitative relaxometry.32 Eighteen of those children participated in the 10-year longitudinal assessments and EBC experiment of study 1 (n = 15 ASD, n = 3 TD). The remainder received MRI with their age 2 diagnosis of ASD (n = 12) or TD (n = 9). Measurements obtained from the structural MRIs were the volumes of the cerebral cortex, cerebellum, hippocampus, and amygdala and the midsagittal area of the anterior lobe vermis, vermis lobules VI-VII, and vermis lobules VIII-X. Midsagittal areas of vermis lobules VI-X were summed to determine the posterior lobe vermis area. Full scanning and morphometry protocols are provided in eMethods (links.lww.com/WNL/C490).

Outcome Measures

Four measures were calculated on each EBC session: the percent of trials with a CR, mean CR onset latency, mean CR peak latency, and mean CR peak amplitude (Figure 1B). Average CR waveforms were calculated for trials in which CR onset and peak occurred within 1 SD of the group mean.

Statistics

Mixed-effects analysis of variance was used to analyze data with repeated measures. Planned comparisons were performed with 2-tailed, 2-sample t tests. Unless noted otherwise, values for trace EBC were means of EBC sessions 1 and 2. Binary and multinomial logistic regression (LR) was performed using the 4 CR measures from each EBC session as independent variables (percent CRs, CR onset and peak latencies, and CR amplitude). The 2 individuals missing data were excluded from LR analyses. Full LR results are presented in eTables 1–8. EBC measure distributions are presented in eFigures 1–2 (links.lww.com/WNL/C490). Analyses were performed using SYSTAT and R software. Statistical significance was p < 0.05. Data are presented as the mean ± SEM, unless noted otherwise.

Data Availability

Data are deposited at the United States National Institutes of Mental Health Data Archive. Further access will be given to qualified investigators on request.

Results

Study 1: EBC Across the Full ASD Spectrum

At age 2 years, children with ASD in the longitudinal cohort exhibited well-below-average intellectual ability (61 ± 2 standard IQ score; 2 ± 0.2 percentile), impaired social communication, social reciprocity, and social responsiveness, and a high degree of restricted interests and repetitive behaviors compared with TD (Figure 2). The ASD cohort was divided into 2 groups based on showing either an improved course of intellectual development or persistent ID (Figure 2, A and B). The threshold for inclusion into the groups was a change in IQ over ages 2–12 years exceeding or not exceeding 1 SD of the normed distribution (15 points). The group showing intellectual development (ASD + noID; n = 16) exhibited significant gains such that by age 11.9 ± 0.2 years, its mean IQ (100 ± 3 standard score; 50 ± 6 percentile; mean IQ increase 2.5 ± 0.2 SD) approached TD peers (120 ± 3 standard score; 82 ± 4 percentile). The group showing persistent ID (ASD + ID, n = 9) exhibited a mean decrease in IQ over 10 years (49 ± 6 standard score at 12.0 ± 0.4 years; 1 ± 0.8 percentile; 0.7 ± 0.2 SD decrease from age 2). At age 11, the ASD + noID group demonstrated improved communication and social functioning and fewer restricted interests and repetitive behaviors (Figure 2, C and D). The sex distribution did not significantly differ between ASD + ID (78% male), ASD + noID (81% male), and TD (71% male).

Figure 2. Longitudinal Characterization of Intellectual Development and Autism Feature Severity for Study 1.

Figure 2

Data in (A, C, and D) are mean ± SEM. Data in (B) are from individual children indicated by sex. (A) IQ scores for the longitudinal groups across ages 2–12 years. (B) Individual IQ scores and 95% confidence intervals at ages 2 and 11 years. The mean IQ for the TD group at age 11 was higher than the normed average (100), but was not unusual for a volunteer TD sample from the UW Autism Center and did not influence the major effects shown by the ASD groups. (C) VABS standard scores for the communication and socialization domains. (D) ASD feature severity for the social affect and restricted interest/repetitive behavior domains as measured by the ADOS. MSEL and DAS-II are age-appropriate assessments of cognitive ability that provide a composite score, commonly referred as IQ, whose distribution has a population mean of 100 and SD of 15. VABS is a semistandardized questionnaire assessment of functional skills in multiple domains, including communication and socialization that provides standard scores having a population mean of 100 and SD of 15. For the MSEL, DAS-II, and VABS tests, lower scores indicate lower performance. The ADOS assesses communication, social interaction, and restricted interests/repetitive behaviors. It was developed to diagnose ASD across a wide range of chronological and mental ages with score ranges reflecting ASD feature severity (1–3, low severity; 4–7, moderate severity; 8–10, high severity). The ADOS provides also a composite severity score reflecting overall ASD severity across a range of 1–10, with 10 being the most severe. ADOS = Autism Diagnostic Observation Schedule; ASD = autism spectrum disorder; TD = typical development; VABS = Vineland Adaptive Behavior Scale.

At age 11 years, the longitudinal cohort was given 4 EBC sessions: 2 sessions of trace EBC, 1 session of long-delay EBC, and 1 session of short-delay EBC. It is generally agreed in both humans and experimental animals that trace EBC elicits a more complex form of associative learning that has a greater dependence on the cerebral cortex and hippocampus than delay EBC14-17 and that the cerebellum facilitates the temporal adaptation of CRs to the CS-US interval.18-21 The latter process importantly involves cerebellar lobule HVI due to its neuroanatomic relation with brain-stem circuits involved in facial nerve control.33 Demonstrating robust temporal adaptation, the onset and peak latencies of the CRs of children with TD acquired to a 700-ms CS-US interval were longer than the latencies of the CRs acquired to a 400-ms CS-US interval (compare Figure 1, B and C with Figure 1D). Thus, participants learn both the CS-US contingency and the duration of the CS-US interval during EBC, which may be distinct neurophysiologic processes.14,15,18,19

The ASD + ID group showed slower learning than TD during trace EBC (p < 0.05; Figure 3A). In addition, the CRs of the ASD + ID group during trace EBC showed abnormally early onset (72 ± 38 ms and 1.4 ± 0.8 SD before TD) and early peak latency (73 ± 41 ms and 1.6 ± 1.0 SD before TD; both p = 0.01; Figure 3B). No significant difference from TD in CR acquisition or CR timing was shown by the ASD + noID group during trace EBC.

Figure 3. Effects of ASD + noID and ASD + ID on Trace and Delay EBC.

Figure 3

Data are mean ± SEM and individuals indicated by sex for the longitudinal ASD groups and TD control group in study 1. (A) Percent CRs in 30-trial blocks across 4 EBC sessions. ASD + ID differed from TD on session 1 of trace EBC (F(1,37) = 6.8, p = 0.013) but on no other session (all p > 0.1). ASD + noID did not differ from TD on any EBC session (all p > 0.16). (B) Mean CR onset latency, CR peak latency, CR amplitude, and CR waveform of the 3 groups during trace EBC. (C) Same as B but for long-delay EBC. (D) Same as B but for short-delay EBC. *p < 0.05 by mixed-effects ANOVA (A) or planned, 2-tailed, 2-sample t test (B–D). Scale = 1× baseline (B–D). ANOVA = analysis of variance; ASD = autism spectrum disorder; ID = intellectual disability; EBC = eye-blink conditioning; TD = typical development

During long-delay EBC, both ASD groups showed abnormally early CR latencies (ASD + noID: 49 ± 20 ms, 0.8 ± 0.3 SD; ASD + ID: 82 ± 42 ms, 1.4 ± 0.7 SD before TD) and abnormally early CR peak latencies (ASD + noID: 46 ± 20 ms, 0.8 ± 0.3 SD; ASD + ID: 82 ± 46 ms, 1.4 ± 0.8 SD before TD; both p < 0.05; Figure 3C). Early-onset CRs in the ASD + noID group during long-delay EBC replicated previous reports.22-24 During short-delay EBC, neither ASD group showed mean percent CRs, CR onset, or CR peak latency that differed more than 0.53 ± 0.5 SD from TD (Figure 3, A and D). The mean CR amplitude of the ASD groups did not differ from TD, except for the ASD + noID group, which showed significantly smaller CRs during trace EBC (p = 0.02; Figure 3B).

Three LR analyses were used to validate the above findings and to help identify a clinical dimension by which EBC measures optimally discriminate ASD subgroups (Figure 4). The first used binary LR to determine whether EBC measures could discriminate ASD from TD. To increase statistical power, we added 7 ASD cases for this cross-sectional analysis (Figure 1A). Using 0 logit as the cutoff, binary LR of EBC measures classified ASD with 81% sensitivity (95% CI 69%–92%), 79% specificity (95% CI 68%–91%), and 80% accuracy (95% CI 72%–88%) (Figure 4A). Receiver operator characteristic analysis produced an area under the curve of 0.87, indicating robust discrimination of ASD from TD (Figure 4B).

Figure 4. LR Analysis of ID and Feature Severity in ASD Using 16 EBC Measures.

Figure 4

Data are individual cases indicating sex (A and D), detection rate (B), and percent correct classification (C). (A) Binary LR analysis showing individuals diagnosed with ASD or TD (dots) vs the logit of the predicted probability of ASD diagnosis based on EBC measures (0 logit = chance probability of ASD or TD diagnosis). A sigmoid function is fit to the categorical data. (B) Receiver operator curve (ROC) of the data in (A) showing 87% area under the curve (AUC), indicating robust classification of ASD and TD by EBC measures. Overall model fit was statistically significant (χ2(16) = 26.7, p = 0.045). (C) Results of 2 multinomial LR analyses showing the percent correct classification of children with ASD into subgroups defined by varying intellectual ability (left) or ASD severity (right) using EBC measures. (D) Scatter plot of ASD severity vs IQ of individuals with ASD over ages 2–11 years. IQ and ASD severity are significantly linearly related in this sample (F(1,130) = 9.26, p = 0.003). ASD = autism spectrum disorder; intellectual disability; EBC = eye-blink conditioning; LR = logistic regression; TD = typical development.

The second analysis used multinomial LR of EBC measures to model the classification of ASD subgroups defined by intellectual development. Three groups for this analysis were the longitudinal groups (ASD + noID, ASD + ID) plus TD controls. The third analysis used multinomial LR of EBC measures and the full cross-sectional cohort to model the classification of subgroups defined by ASD severity. Three groups for the third analysis were defined by the ADOS adjectival cutoffs: high severity (ADOS severity score 8–10), low-to-moderate severity (ADOS severity score 1–7), and TD (no diagnosis).

Multinomial LR using EBC measures most accurately classified children with ASD along the dimension of intellectual development (94% accuracy; 95% CI 89%–100%) with sensitivities of 93% for TD (95% CI 85%–100%), 94% for ASD + noID (95% CI 83%–100%), and 100% for ASD + ID. Multinomial LR less accurately classified children along the dimension of ASD severity (85% accuracy, 95% CI 77%–93%) with sensitivities of 93% for TD (95% CI 85%–100%), 67% for low-to-moderate severity ASD (95% CI 52%–81%), and 92% for high-severity ASD (95% CI 79%–100%) (Figure 4C).

EBC measures with significant odds ratios for discriminating children with ASD based on intellectual development validated the measures that were significant by analyses of central tendency (Figure 3), including percent CRs on session 1 of trace EBC (3.47, 95% CI 1.54–7.83, p = 0.003), CR peak latency on session 2 trace EBC (0.8, 95% CI 0.68–0.94, p = 0.008), and CR peak latency during long-delay EBC (1.21, 95% CI 1.02–1.45, p = 0.034), but also CR amplitude during long-delay EBC (560.26, 95% CI 430.06–729.88, p < 0.001). EBC measures with significant odds ratios for classifying children based on ASD severity overlapped with those that were significant for classification based on intellectual development, such as percent CRs on session 1 of trace EBC (0.73, 95% CI 0.56–0.96, p = 0.026) and CR peak latency during long-delay EBC (1.78, 95% CI 1.03–3.05, p = 0.038), but included also CR onset latency during long-delay EBC (0.55, 95% CI 0.31–0.98, p = 0.042) and CR amplitude on session 2 of trace EBC (0.35, 95% CI 0.14–0.90, p = 0.029).

In summary, study 1 demonstrated that EBC measures most accurately resolved the heterogeneity of ASD along the dimension of intellectual development. EBC measures that distinguished ASD + ID were specific to trace EBC, whereas EBC measures during long-delay EBC distinguished ASD independent of ID. Classification accuracy by LR modeling using EBC measures was less strong when ASD subgroups were defined by ASD severity, although remained significant. This was likely due to significant covariation of ASD severity and IQ in our sample (r = −0.26, p = 0.003; Figure 4D). Thus, the ability of EBC measures to classify children along the dimension of ASD severity may be due to an influence of ID on ASD features.

Study 2: Age 2 Brain Morphometry

We divided the 15 children with ASD in study 1 who underwent MRI at age 2 into 3 subgroups: (1) those with early CR onsets during trace EBC (ASD–early trace, n = 7); (2) those with early CR onsets only during long-delay EBC (ASD–early long-delay, n = 4); and (3) those not showing a decrease in CR onset in either paradigm (ASD–no change, n = 4). Inclusion criteria for the 3 groups, respectively, were (1) CR onset latency during trace EBC earlier than 0.5 SD below the TD mean, (2) not meeting criterion 1 but having CR onset latency during long-delay EBC below the TD mean, and (3) not meeting either criteria. We hypothesized that the CR timing impairments with ASD identified in study 1 might be related to different patterns and magnitudes of brain hypoplasia.

Mean CR waveforms demonstrated the differences between the ASD–early trace and ASD–early long-delay groups (Figure 5, A and B). The mean CR onset of the ASD–early trace group was 111 ± 37 ms (2.2 ± 0.8 SD) before TD during trace EBC and 156 ± 36 ms (2.6 ± 0.6 SD) before TD during long-delay EBC. The mean CR onset of the ASD–early long-delay group was 23 ± 14 ms (0.5 ± 0.3 SD) after TD during trace EBC and 66 ± 35 ms (1.1 ± 0.6 SD) before TD during long-delay EBC. The mean CR onset of the ASD–no change group was 38 ± 8 ms (0.8 ± 0.2 SD) after TD during trace EBC and 39 ± 12 ms (0.6 ± 0.2 SD) after TD during long-delay EBC.

Figure 5. Brain Structure Volumes at Age 2 in Relation to Age 11 EBC.

Figure 5

Data are mean ± SEM and individuals indicated by sex. (A and B) Mean CR waveforms for study 2 participants with ASD during trace (A) and long-delay (B) EBC compared with TD. The participants with ASD are grouped by having early-onset CRs during trace EBC (purple, n = 7) or early-onset CRs only during long-delay EBC (green, n = 4). (C) Mean CR waveform during short-delay EBC of a negative control group comprising children with ASD having early-onset CRs (brown, n = 8) compared with TD. (D) Age 2 volumes of the cerebellum, cerebral cortex, hippocampus, and amygdala for the ASD groups in (A–C) compared with TD. *p < 0.05 vs TD by planned, 2-tailed, 2-sample t test. Scale = 1× baseline (A–C). ASD = autism spectrum disorder; EBC = eye-blink conditioning; TD = typical development.

Cerebellar volume in the ASD–early trace group was significantly less than TD (118 ± 4 vs 129 ± 4 cm3, 9 ± 3% decrease, 0.92 ± 0.31 SD below TD, p = 0.048; Figure 5D). The reduction in cerebellar volume in the ASD–early trace group was not accompanied by significant changes in the volume of the cerebral cortex, hippocampus, or amygdala (all p ≥ 0.5). Cerebellar volumes of the ASD–early long-delay (133 ± 3 cm3) and ASD–no change (132 ± 3 cm3) groups did not differ from TD (p = 0.5), as did no other brain area (all p ≥ 0.2, Figure 5D).

As a negative control, we defined an artificial phenotype of early-onset CRs during short-delay EBC that was not shown by children with ASD (Figure 5C). The criterion for inclusion into this ASD-negative control group was the same used to identify early-onset CRs during trace EBC, but for short-delay EBC. The mean CR onset latency of the ASD negative control group (n = 8) during short-delay EBC was 45 ± 11 ms (1.4 ± 0.3 SD) before TD. Early-onset CRs during short-delay EBC in children with ASD were not accompanied by significant reductions in the volume of the cerebellum, cerebral cortex, hippocampus, or amygdala (all p ≥ 0.3, Figure 5D).

We tested whether changes in the size of the cerebellar anterior and posterior lobes contributed to cerebellar hypoplasia in the ASD–early trace group (Figure 6A). A significant 21 ± 3% reduction in the midsagittal area of the posterior lobe vermis was found for the ASD–early trace group (p = 0.03; Figure 6B). This change was due to a significant 21 ± 4% reduction in vermis lobules VIII-X (p = 0.02) and a trend level, 20 ± 3% reduction in vermis lobules VI-VII (p = 0.06). The reduction was specific to the posterior lobe, as the anterior lobe vermis did not differ from TD (14 ± 4% reduction, p = 0.1). Neither the ASD–early long-delay nor the ASD–no change group showed significant reductions in the posterior lobe vermis (0.6 ± 5% and 9 ± 5% decrease, respectively) or anterior lobe vermis (1 ± 4% increase, 6 ± 9% decrease, respectively) compared with TD (all p ≥ 0.8). In addition, the negative ASD control group with early-onset CRs during short-delay EBC showed no significant decrease in the midsagittal area of any vermis region (all p ≥ 0.09, Figure 6B).

Figure 6. Age 2 Vermis Hypoplasia and CR Timing.

Figure 6

Data are mean +SEM and individuals indicated by sex. (A) Midsagittal MRI of the cerebellum at age 2 years indicating the vermis lobules studied. (B) Midsagittal vermis area for children with TD (black) and ASD with early-onset CRs only during long-delay EBC (green) or early-onset CRs during trace EBC (purple). Children with ASD with early-onset CRs during trace EBC showed significant reductions in the midsagittal area of the posterior lobe vermis and lobules VIII-X. Children with ASD with early-onset CRs only during long-delay EBC did not show vermis hypoplasia. Identical analysis of the negative control group defined by early-onset CRs during short-delay EBC (brown) did not show vermis hypoplasia. *p < 0.05 by planned, 2-tailed, 2-sample t test. ASD = autism spectrum disorder; TD = typical development.

To overcome limitations of the small sample size, we combined the MRI cohort from study 1 with a second cohort of children with ASD (n = 12) or TD (n = 9) that also had structural brain imaging at age 2 and diagnostic assessments at ages 2, 3, and 4 years. With the combined cohort, we tested whether cerebellar hypoplasia of the magnitude implicated in study 1 prospectively predicted ID and more severe ASD. Cerebellar hypoplasia was defined as being smaller than 1 SD below the TD mean in 1 of 4 measures: cerebellar volume, cerebellar volume normalized by cerebral cortex volume, midsagittal area of the posterior lobe vermis, and midsagittal area of the posterior lobe vermis normalized by midsagittal area of the anterior lobe vermis. By each measure, we identified 7–11 children with ASD with cerebellar hypoplasia at age 2.

The means of the ASD + cerebellar hypoplasia groups relative to TD were cerebellar volume, 1.48 ± 0.16 SD smaller; normalized cerebellar volume, 1.55 ± 0.10 SD smaller; posterior lobe vermis area, 1.29 ± 0.04 SD smaller; and normalized posterior lobe vermis area, 1.21 ± 0.04 SD smaller. Children with ASD + cerebellar hypoplasia were compared with children with ASD + cerebellar normoplasia (the means relative to TD were cerebellar volume, 0.16 ± 0.12 SD larger; normalized cerebellar volume, 0.07 ± 0.22 SD smaller; posterior lobe vermis area, 0.30 ± 0.16 SD smaller; and normalized posterior lobe vermis area, 0.14 ± 0.17 SD smaller).

There was no significant difference in intellectual development over ages 2–4 due to cerebellar hypoplasia defined by any of the 4 measures (all p ≥ 0.2). Negative outcomes also were found for ASD severity over ages 2–4 for 3 of the measures. When cerebellar hypoplasia was defined by reduced normalized posterior lobe vermis area, children with ASD + cerebellar hypoplasia showed greater ASD severity of modest clinical significance compared with children with ASD + cerebellar normoplasia (7.9 ± 0.3 vs 6.6 ± 0.3 ADOS composite score; p = 0.02). Figure 7A shows the ASD group defined by the presence of posterior lobe vermis hypoplasia at a magnitude slightly greater than that identified in study 1 (Figure 6B). Notably, this measure of cerebellar hypoplasia did not result in greater ID or ASD severity (Figure 7B). Longitudinal diagnostic measures pertaining to all 4 measures of cerebellar hypoplasia are provided in eFigure 3 (links.lww.com/WNL/C490).

Figure 7. Relationships Between Age 2 Vermis Hypoplasia and Early-Onset CRs to Intellectual Development and ASD Severity.

Figure 7

Data are mean ± SEM and individuals indicated by sex. (A and B) Prospective longitudinal study of intellectual development and ASD severity following MRI finding of posterior lobe vermis hypoplasia at age 2 years. (A) Midsagittal area of the posterior lobe vermis at age 2 in children with TD (black bar, n = 9), ASD + vermis normoplasia (gray bar, n = 20), and ASD + vermis hypoplasia (open bar, n = 7). Hypoplasia was defined as being smaller than 1 SD below the TD mean. (B) Longitudinal analysis of IQ (solid lines) and ASD severity (dashed lines) over ages 2–4 years for the 2 ASD groups in (A) (matching colors). The 2 ASD groups did not differ statistically in intellectual development or ASD severity over ages 2–4 years (both p ≥ 0.7). Diagnostic measures for TD are presented at age 2. (C and D) Retrospective analysis of intellectual development and ASD severity in groups studied with EBC at age 11. (C) Mean CR waveforms during trace EBC for TD (black, n = 6) and for ASD groups defined by early-onset CRs during trace EBC (purple, n = 5) or early-onset CRs only during long-delay EBC (green, n = 5). Scale = 1× baseline. (D) Longitudinal analysis of IQ (solid lines) and ASD severity (dashed lines) over ages 2–4 for the ASD groups in (C) (matching colors). The ASD groups differed significantly in IQ (significant main effect of group F(1,8) = 6.8, p = 0.03 and interaction of group by age F(2,16) = 6.9, p = 0.007) but not in ASD severity (main effect of group and interaction of group by age both p > 0.1). Early-latency CRs during trace EBC were retrospectively associated with greatly impaired intellectual development. ASD = autism spectrum disorder; EBC = eye-blink conditioning; TD = typical development.

Finally, to better understand the relationship between ASD severity and CR timing, we tested whether trace EBC had special relevance for identifying ID and severe ASD. We compared 2 subgroups of children with ASD from study 1 (n = 5 each) having large reductions in CR onset latency exceeding 1.5 SD of the TD mean during trace EBC (purple, Figure 7C) or only during long-delay EBC (green, Figure 7C). Children with ASD with early-onset CRs during trace EBC at age 11 showed no measurable intellectual development over ages 2–4 years compared with children with ASD and early-onset CRs only during long-delay EBC (p = 0.007; Figure 7D). There was no significant difference in ASD severity between the 2 ASD groups as measured by the ADOS (p = 0.5, Figure 7D).

Discussion

We demonstrated that EBC learning paradigms can help resolve the heterogeneity of intellectual development and feature severity in ASD and provide a behavioral marker linked to cerebellar hypoplasia. In our prospective analysis, cerebellar hypoplasia per se at age 2 years did not predict greater ID over ages 2–4 in children with ASD, although 1 measure of posterior lobe vermis hypoplasia predicted slightly greater ASD severity. More robustly, our retrospective analysis at age 11 demonstrated that slow learning and early-onset CRs during trace EBC were associated both with persistent childhood ID and prominent hypoplasia of the cerebellum at age 2 years. Because early-onset CRs are consistent with cerebellar cortex pathophysiology,20,21 the findings support the hypothesis of a cerebellar contribution to ASD + ID and covarying ASD features that present in some children as cerebellar hypoplasia at age 2 years.

The findings help explain the significance of cerebellar hypoplasia in ASD, an example of brain dysmorphology that helped establish ASD as a neurodevelopmental disorder.3,4,13 The finding of cerebellar hypoplasia in ASD has been replicated34-38 and is associated with the loss of cerebellar Purkinje cell somata and myelinated axons in the posterior lobe that contribute to reduced cerebellar volume.39,40 Nevertheless, not all studies replicated the original observation and those null outcomes9-12 have been reinforced by uncertainty of how the cellular properties and network physiology of the cerebellum—well documented for motricity41,42—may contribute to intellect or contribute to the nonmotor domains affected by ASD.

Early-onset CRs during long-delay EBC in children with ASD + noID were previously reported by 2 research groups.22-24 Our finding of early-onset CRs during long-delay EBC in children with ASD + noID and children with ASD + ID demonstrates that this phenotype is a general characteristic of ASD. However, impairments in CR timing during long-delay EBC are not directly associated with early-childhood cerebellar hypoplasia and do not help explain ID or the heterogeneity of ASD severity. Thus, long-delay EBC by itself cannot be used to assess the heterogeneity of brain mechanisms responsible for the ASD spectrum.

By studying children across the full ASD spectrum whose heterogeneity in ASD severity and intellectual development was documented over 10 years, we demonstrated that slower associative learning and impairments in CR timing—specifically during trace EBC—are related to persistent ID and covarying ASD features. Our findings have 3 important implications for understanding the pathophysiology of ASD and cerebellar functions. First, the findings demonstrate that trace EBC interrogates the brain circuit pathophysiology that contributes to ASD + ID. As such, trace EBC offers an advantage over paradigms that test cerebellar functions that are not affected by ASD,43 may not be sufficiently sensitive to assess brain pathophysiology involved in ID, or may not be amenable for testing very severely affected children.38 Second, the findings indicate the likelihood of differential involvement of the cerebellum in learning and CR timing during trace vs delay EBC. Although cerebellar lesion studies have specifically implicated the posterior lobe HVI for the adaptive timing of CRs, our study suggests that ASD + ID involves additional cerebellar lobules as evidenced by the magnitude of cerebellar hypoplasia that co-occurs with altered CR timing during trace EBC. Although we did not measure HVI volume, HVI alone is unlikely to account for the 9% reduction in total cerebellar volume observed here. Moreover, the 21% reduction in posterior lobe vermis associated with early-onset CRs during trace EBC indicates that cerebellar hypoplasia in children with ASD + ID likely involves posterior lobe lobules that are not directly associated with eye-blink motricity.33 Indeed, current understanding of the large posterior lobe lobules HVII and HVIII indicates that they contribute to nonmotor functioning via their trans-synaptic connections to the nonmotor thalamus and prefrontal cortex.44 Third, the findings emphasize that cerebellar hypoplasia in children with ASD is not sufficient to produce severe ID, although children with ASD plus early-onset CRs during trace EBC showed reduced cerebellar volume and posterior lobe vermis area. We conclude that cerebellar hypoplasia contributes to ID and covarying ASD features only in conjunction with pathophysiology elsewhere in the brain.

What contribution might the cerebellum make to intellectual development? A testable hypothesis is that the cerebellum promotes intellectual development by facilitating implicit timing. Implicit timing is the subconscious use of time intervals that facilitates temporal prediction, sensory processing, prosody of speech, and movement.45,46 For example, experience with a subsecond time interval bounded by 2 stimuli promotes anticipation of the second stimulus that is expressed subconsciously as a cerebellar-dependent, anticipatory, perceptual response in the cerebral cortex and an anticipatory motor response.47 The temporal adaptation of CRs to the CS-US interval during EBC is another example of implicit timing that is also cerebellar dependent.18-21 As measured by fMRI, trace EBC coactivated the posterior cerebellum, prefrontal cortex, and hippocampus to a greater extent than delay EBC,14 revealing a structure of cerebrocerebellar interaction that overlaps with brain circuitries involved in intellect. We therefore hypothesize that impairments in implicit timing contribute to ASD + ID following cerebellar pathophysiology due to diaschetic hypofunction in forebrain systems traditionally associated with learning and intellect.

The 81% sensitivity of EBC for differentiating children with ASD from TD approximates the 88% sensitivity of an assessment of ASD likelihood based on gray matter measurements in the neocortex.2 Relevant to our study was that surface area expansion of the middle frontal gyrus showed the largest difference between children at low vs high familial likelihood of ASD,2 within which lie Brodmann areas 9 and 46 that are the primary receiving zones of trans-synaptic input from posterior lobe cerebellar lobules44 that are activated during cognition.6 The results suggest that physiologic brain imaging during trace EBC will elucidate the cerebellar contribution to intellect.

Standard clinical assessments at age 2 do not predict intellectual development with high levels of certainty. However, combining well-understood functional tests such as EBC with brain morphology in young children may enhance our ability to predict intellectual development. Replication studies should overcome this study's limitations by using a prospective analysis of intellectual development using coincident, early-childhood EBC, higher-resolution morphometry, and larger cohorts than were available to us to allow for both training and validation of the logistic models for ASD participant characterization.

Acknowledgment

The authors especially thank the participating families for their generosity and unwavering commitment to brain research. They thank Dr. J. Greenson for consultation and referrals into the study, B.F. Sparks, R. N. and Dr. D.W.W. Shaw for assistance in analyzing the brain anatomy, L. Tsui and C. Sinquimani for assisting with the participants, and L. Tsui, J. Raduazzo, and S. Kamm for participant coordination. They gratefully acknowledge Drs. G. Dawson and S. Rogers for initiating the longitudinal study.

Glossary

ASD

autism spectrum disorder

CS

conditioned stimulus

DAS

Differential Abilities Scale

EBC

eye-blink conditioning

ID

intellectual disability

LR

logistic regression

MSEL

Mullen Scales of Early Learning

TD

typical development

US

unconditioned stimulus

Appendix. Authors

Appendix.

Study Funding

This work was supported by a grant from the United States National Institute of Mental Health (R01 MH100887 to JPW). The diagnostic and brain imaging data of the longitudinal cohort over ages 2–6 were acquired at the University of Washington Autism Center and Diagnostic Imaging Sciences Center with the support of the United States National Institutes of Mental Health (U54 MH066399). Many of the assessments were conducted at the University of Washington Center for Human Development and Disability, supported by the United States Eunice Kennedy Shriver National Institute of Child Health & Human Development (U54 HD083091). The funders of the study had no role in the design of the study, data collection, data analysis, data interpretation, or the writing of the report.

Disclosure

The authors report no disclosures relevant to the manuscript. Go to Neurology.org/N for full disclosures.

References

  • 1.Dawson G, Meltzoff AN, Osterling J, Rinaldi J. Neuropsychological correlates of early symptoms of autism. Child Develop. 2008;69(5):1276-1285. doi: 10.1111/j.1467-8624.1998.tb06211.x [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 2.Hazlett HC, Gu H, Munsell BC, et al. Early brain development in infants at high risk for autism spectrum disorder. Nature. 2017;542(7641):348-351. doi: 10.1038/nature21369 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3.Courchesne E, Hesselink JR, Jernigan TL, Yeung-Courchesne R. Abnormal neuroanatomy in a nonretarded person with autism. Unusual findings with magnetic resonance imaging. Arch Neurol. 1987;44(3):335-341. doi: 10.1001/archneur.1987.00520150073028 [DOI] [PubMed] [Google Scholar]
  • 4.Courchesne E, Yeung-Courchesne R, Hesselink J, Jernigan T, Jernigan TL. Hypoplasia of cerebellar vermal lobules VI and VII in autism. N Engl J Med. 1988;318(21):1349-1354. doi: 10.1056/nejm198805263182102 [DOI] [PubMed] [Google Scholar]
  • 5.Stoodley CJ, MacMore JP, Makris N, Sherman JC, Schmahmann JD. Location of lesion determines motor vs. cognitive consequences in patients with cerebellar stroke. NeuroImage Clin. 2016;12:765-775. doi: 10.1016/j.nicl.2016.10.013 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Stoodley CJ, Schmahmann JD. Functional topography in the human cerebellum: a meta-analysis of neuroimaging studies. Neuroimage. 2009;44(2):489-501. doi: 10.1016/j.neuroimage.2008.08.039 [DOI] [PubMed] [Google Scholar]
  • 7.Wong CHY, Liu J, Lee TMC, et al. Fronto-cerebellar connectivity mediating cognitive processing speed. Neuroimage. 2021;226:117556. doi: 10.1016/j.neuroimage.2020.117556 [DOI] [PubMed] [Google Scholar]
  • 8.Dave S, VanHaerents S, Voss JL. Cerebellar theta and beta noninvasive stimulation rhythms differentially influence episodic memory versus semantic prediction. J Neurosci. 2020;40(38):7300-7310. doi: 10.1523/jneurosci.0595-20.2020 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Piven J, Nehme E, Simon J, Barta P, Pearlson G, Folstein SE. Magnetic resonance imaging in autism: measurement of the cerebellum, pons, and fourth ventricle. Biol Psychiatry. 1992;31(5):491-504. doi: 10.1016/0006-3223(92)90260-7 [DOI] [PubMed] [Google Scholar]
  • 10.Piven J, Saliba K, Bailey J, Arndt S. An MRI study of autism: the cerebellum revisited. Neurology. 1997;49(2):546-551. doi: 10.1212/wnl.49.2.546 [DOI] [PubMed] [Google Scholar]
  • 11.Manes F, Piven J, Vrancic D, Nanclares V, Plebst C, Starkstein SE. An MRI study of the corpus callosum and cerebellum in mentally retarded autistic individuals. J Neuropsychiatr Clin Neurosci. 1999;11(4):470-474. doi: 10.1176/jnp.11.4.470 [DOI] [PubMed] [Google Scholar]
  • 12.Scott JA, Schumann CM, Goodlin-Jones BL, Amaral DG. A comprehensive volumetric analysis of the cerebellum in children and adolescents with autism spectrum disorder. Autism Res. 2009;2(5):246-257. doi: 10.1002/aur.97 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Fatemi SH, Aldinger KA, Ashwood P, et al. Consensus paper: pathological role of the cerebellum in autism. Cerebellum. 2012;11(3):777-807. doi: 10.1007/s12311-012-0355-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Cheng DT, Disterhoft JF, Power JM, Ellis DA, Desmond JE. Neural substrates underlying human delay and trace eyeblink conditioning. Proc Natl Acad Sci USA. 2008;105(23):8108-8113. doi: 10.1073/pnas.0800374105 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Weiss C, Disterhoft JF. Exploring prefrontal cortical memory mechanisms with eyeblink conditioning. Behav Neurosci. 2011;125(3):318-326. doi: 10.1037/a0023520 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16.Clark RE, Manns JR, Squire LR. Classical conditioning, awareness, and brain systems. Trends Cogn Sci. 2002;6(12):524-531. doi: 10.1016/s1364-6613(02)02041-7 [DOI] [PubMed] [Google Scholar]
  • 17.Siegel JJ, Taylor W, Gray R, et al. Trace eyeblink conditioning in mice is dependent upon the dorsal medial prefrontal cortex, cerebellum, and amygdala: behavioral characterization and functional circuitry. eNeuro. 2015;2:ENEURO.0051-14.2015. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18.Welsh JP. Changes in the motor pattern of learned and unlearned responses following cerebellar lesions: a kinematic analysis of the nictitating membrane reflex. Neuroscience. 1992;47:1-19. doi: 10.1016/0306-4522(92)90116-j [DOI] [PubMed] [Google Scholar]
  • 19.Welsh JP, Harvey JA. Cerebellar lesions and the nictitating membrane reflex: performance deficits of the conditioned and unconditioned response. J Neurosci. 1989;9(1):299-311. doi: 10.1523/jneurosci.09-01-00299.1989 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Kalmbach BE, Davis T, Ohyama T, Riusech F, Nores WL, Mauk MD. Cerebellar cortex contributions to the expression and timing of conditioned eyelid responses. J Neurophysiol. 2010;103(4):2039-2049. doi: 10.1152/jn.00033.2010 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21.Gerwig M, Kolb FP, Timmann D. The involvement of the human cerebellum in eyeblink conditioning. Cerebellum. 2007;6(1):38-57. doi: 10.1080/14734220701225904 [DOI] [PubMed] [Google Scholar]
  • 22.Sears LL, Finn PR, Steinmetz JE. Abnormal classical eye-blink conditioning in autism. J Autism Dev Disord. 1994;24(6):737-751. doi: 10.1007/bf02172283 [DOI] [PubMed] [Google Scholar]
  • 23.Oristaglio J, Hyman West S, Ghaffari M, et al. Children with autism spectrum disorders show abnormal conditioned response timing on delay, but not trace, eyeblink conditioning. Neuroscience. 2013;248:708-718. doi: 10.1016/j.neuroscience.2013.06.007 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24.Welsh JP, Oristaglio JT. Autism and classical eyeblink conditioning: performance changes of the conditioned response related to autism spectrum disorder diagnosis. Front Psychiatry. 2016;7:137. doi: 10.3389/fpsyt.2016.00137 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25.Dawson G, Rogers S, Munson J, et al. Randomized, controlled trial of an intervention for toddlers with autism: the Early Start Denver Model. Pediatrics. 2010;125(1):e17-e23. doi: 10.1542/peds.2009-0958 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26.Estes A, Munson J, Rogers SJ, Greenson J, Winter J, Dawson G. Long-term outcomes of early intervention in 6-year-old children with autism spectrum disorder. J Am Acad Child Adolesc Psychiatry. 2015;54(7):580-587. doi: 10.1016/j.jaac.2015.04.005 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27.Mullen EM. Mullen Scales of Early Learning: AGS Edition. American Guidance Service Publishing; 1995. [Google Scholar]
  • 28.Canivez GL, McGill RJ. Factor structure of the differential ability scales–second edition: exploratory and hierarchical factor analyses with the core subtests. Psychol Assess. 2016;28:1475-1488. [DOI] [PubMed] [Google Scholar]
  • 29.Lord C, Rutter M, DiLavore PC, Risi S. Autism Diagnostic Observation Schedule. Western Psychological Services; 2000. [Google Scholar]
  • 30.Sparrow S, Balla D, Cicchetti D. Vineland Adaptive Behavior Scales: Second Edition. American Guidance Service; 2005. [Google Scholar]
  • 31.Gormezano I. Classical conditioning. In: Sidowski JB, editor. Experimental Methods and Instrumentation in Psychology. McGraw-Hill; 1966:385-420. [Google Scholar]
  • 32.Petropoulos H, Friedman SD, Shaw DW, Artru AA, Dawson G, Dager SR. Gray matter abnormalities in autism spectrum disorder revealed by T2 relaxation. Neurology. 2006;67(4):632-636. doi: 10.1212/01.wnl.0000229923.08213.1e [DOI] [PubMed] [Google Scholar]
  • 33.Gonzalez-Joekes J, Schreurs BG. Anatomical characterization of a rabbit cerebellar eyeblink premotor pathway using pseudorabies and identification of a local modulatory network in anterior interpositus. J Neurosci. 2012;32(36):12472-12487. doi: 10.1523/JNEUROSCI.2088-12.2012 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34.Courchesne E, Townsend J, Saitoh O. The brain in infantile autism: posterior fossa structures are abnormal. Neurology. 1994;44(2):214-223. doi: 10.1212/wnl.44.2.214 [DOI] [PubMed] [Google Scholar]
  • 35.Murakami JW, Courchesne E, Press GA, Yeung-Courchesne R, Hesselink JR. Reduced cerebellar hemisphere size and its relationship to vermal hypoplasia in autism. Arch Neurol. 1989;46(6):689-694. doi: 10.1001/archneur.1989.00520420111032 [DOI] [PubMed] [Google Scholar]
  • 36.Ciesielski KT, Harris RJ, Hart BL, Pabst HF. Cerebellar hypoplasia and frontal lobe cognitive deficits in disorders of early childhood. Neuropsychologia. 1997;35(5):643-655. doi: 10.1016/s0028-3932(96)00119-4 [DOI] [PubMed] [Google Scholar]
  • 37.Webb SJ, Sparks BF, Friedman SD, et al. Cerebellar vermal volumes and behavioral correlates in children with autism spectrum disorder. Psychiatry Res Neuroimaging. 2009;172(1):61-67. doi: 10.1016/j.pscychresns.2008.06.001 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 38.McKinney WS, Kelly SE, Unruh KE, et al. Corrigendum: cerebellar volumes and sensorimotor behavior in autism spectrum disorder. Front Integr Neurosci. 2022;16:1020980. doi: 10.3389/fnint.2022.1020980 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 39.Ritvo ER, Freeman BJ, Scheibel AB, et al. Lower Purkinje cell counts in the cerebella of four autistic subjects: initial findings of the UCLA-NSAC autopsy research report. Am J Psychiatry. 1986;143(7):862-866. doi: 10.1176/ajp.143.7.862 [DOI] [PubMed] [Google Scholar]
  • 40.Whitney ER, Kemper TL, Rosene DL, Bauman ML, Blatt GJ. Density of cerebellar basket and stellate cells in autism: evidence for a late developmental loss of Purkinje cells. J Neurosci Res. 2009;87(10):2245-2254. doi: 10.1002/jnr.22056 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 41.Welsh JP, Llinás RR. Some organizing principles for the control of movement based on olivocerebellar physiology. Prog Brain Res. 1997;114:449-461. [DOI] [PubMed] [Google Scholar]
  • 42.Welsh JP, Turecek J. Implications of electrical synapse plasticity in the inferior olive. In: Jing J, editor. Network Functions and Plasticity: Perspectives from Studying Neuronal Electrical Coupling in Microcircuits. Academic Press; 2017:262-302. [Google Scholar]
  • 43.Goldberg MC, Landa R, Lasker A, Cooper L, Zee DS. Evidence of normal cerebellar control of the vestibulo-ocular reflex (VOR) in children with high-functioning autism. J Autism Develop Disord. 2000;30(6):519-524. doi: 10.1023/a:1005631225367 [DOI] [PubMed] [Google Scholar]
  • 44.Middleton FA, Strick PL. Cerebellar projections to the prefrontal cortex of the primate. J Neurosci. 2001;21(2):700-712. doi: 10.1523/jneurosci.21-02-00700.2001 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 45.Coull JT, Nobre AC. Dissociating explicit timing from temporal expectation with fMRI. Curr Opin Neurobiol. 2008;18(2):137-144. doi: 10.1016/j.conb.2008.07.011 [DOI] [PubMed] [Google Scholar]
  • 46.Breska A, Ivry RB. Taxonomies of timing: where does the cerebellum fit in?. Curr Opin Behav Sci. 2016;8:282-288. doi: 10.1016/j.cobeha.2016.02.034 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 47.Trillenberg P, Verleger R, Teetzmann A, Wascher E, Wessel K. On the role of the cerebellum in exploiting temporal contingencies: evidence from response times and preparatory EEG potentials in patients with cerebellar atrophy. Neuropsychologia. 2004;42(6):754-763. doi: 10.1016/j.neuropsychologia.2003.11.005 [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Data Availability Statement

Data are deposited at the United States National Institutes of Mental Health Data Archive. Further access will be given to qualified investigators on request.


Articles from Neurology are provided here courtesy of American Academy of Neurology

RESOURCES