Prediction learning in adults with autism and its molecular correlates

Abstract

Background According to Bayesian hypotheses, individuals with Autism Spectrum Disorder (ASD) have difficulties making accurate predictions about their environment. In particular, the mechanisms by which they assign precision to predictions or sensory inputs would be suboptimal in ASD. These mechanisms are thought to be mostly mediated by glutamate and GABA. Here, we aimed to shed light on prediction learning in ASD and on its neurobiological correlates. Methods Twenty-six neurotypical and 26 autistic adults participated in an associative learning task where they had to learn a probabilistic association between a tone and the rotation direction of two dots, in a volatile context. They also took part in magnetic resonance spectroscopy (MRS) measurements to quantify Glx (glutamate and glutamine), GABA + and glutathione in a low-level perceptual region (occipital cortex) and in a higher-level region involved in prediction learning (inferior frontal gyrus). Results Neurotypical and autistic adults had their percepts biased by their expectations, and this bias was smaller for individuals with a more atypical sensory sensitivity. Both groups were able to learn the association and to update their beliefs after a change in contingency. Interestingly, the percentage of correct predictions was correlated with the Glx/GABA + ratio in the occipital cortex (positive correlation) and in the right inferior frontal gyrus (negative correlation). In this region, MRS results also showed an increased concentration of Glx in the ASD group compared to the neurotypical group. Limitations We used a quite restrictive approach to select the MR spectra showing a good fit, which led to the exclusion of some MRS datasets and therefore to the reduction of the sample size for certain metabolites/regions. Conclusions Autistic adults appeared to have intact abilities to make predictions in this task, in contrast with the Bayesian hypotheses of ASD. Yet, higher ratios of Glx/GABA + in a frontal region were associated with decreased predictive abilities, and ASD individuals tended to have more Glx in this region. This neurobiological difference might contribute to suboptimal predictive mechanisms in ASD in certain contexts.

Increased prefrontal GABA concentrations in adults with autism spectrum disorders

“…MRS studies in adults with ASD reported no significant GABA concentration differences (Ajram et al, 2017; Horder et al, 2018; Kirkovski et al, 2018) or increased GABA levels (Fung et al, 2021) in frontal and prefrontal areas compared to NTC. With regard to Glu or Glx concentrations, increased (Sapey‐Triomphe et al, 2021) as well as similar Glu or Glx concentrations (Ajram et al, 2017; Endres et al, 2015; Horder et al, 2013, 2018) have been reported in the frontal or prefrontal cortex of adults with ASD.…”

“…According to the E/I imbalance theory, this imbalance between excitatory and inhibitory neurotransmission is reflected in an increase in glutamatergic (excitatory) and a decrease in GABAergic (inhibitory) signaling (Rubenstein, 2010; Rubenstein & Merzenich, 2003). However, with respect to glutamate, both a hypoglutamatergic and a hyperglutamatergic theory has been proposed (Eltokhi et al, 2020; Sapey‐Triomphe et al, 2021). Based on animal studies in which hypoglutamatergic animals exhibited autistic behavior, the hypoglutamatergic hypothesis of ASD has been postulated (Carlsson, 1998).…”

“…MRS investigations detected increased Glu or Glx concentrations (Adults: amygdala‐hippocampal complex (Page, 2006), inferior frontal gyrus (Sapey‐Triomphe et al, 2021), and auditory cortex (Brown et al, 2013); children: sensorimotor cortex (He et al, 2021), ACC (Bejjani et al, 2012; Hassan et al, 2013; Joshi et al, 2013; Naaijen et al, 2017), frontal lobe (Hassan et al, 2013), and putamen (Doyle‐Thomas et al, 2014)), while other studies found decreased levels (Adults: ACC (Bernardi et al, 2011; Tebartz van Elst et al, 2014a, 2014b), basal ganglia (Horder et al, 2013; Horder et al, 2018); children: cerebellum (DeVito et al, 2007), frontal lobe (Kubas et al, 2012), thalamus (Hegarty et al, 2018)) or no significant differences (e.g. Adults: frontal regions (Ajram et al, 2017; Endres et al, 2015; Horder et al, 2013, 2018), ACC (Endres et al, 2017; Libero et al, 2015, 2016), thalamus (Bernardi et al, 2011); children: medial prefrontal cortex (Carvalho Pereira et al, 2018), ACC (Brix et al, 2015; Cochran et al, 2015; Goji et al, 2017; Ito et al, 2017), caudate or thalamus (Doyle‐Thomas et al, 2014)).…”

“…Adults: frontal regions (Ajram et al, 2017; Endres et al, 2015; Horder et al, 2013, 2018), ACC (Endres et al, 2017; Libero et al, 2015, 2016), thalamus (Bernardi et al, 2011); children: medial prefrontal cortex (Carvalho Pereira et al, 2018), ACC (Brix et al, 2015; Cochran et al, 2015; Goji et al, 2017; Ito et al, 2017), caudate or thalamus (Doyle‐Thomas et al, 2014)). A recent investigation in adults with ASD reported elevated ratios of Glx/GABA+ in the right inferior frontal gyrus being associated with reduced predictive abilities as well as increased Glx levels in this region (Sapey‐Triomphe et al, 2021). The variability of previous findings may be due to differences in study samples (e.g., age groups, inclusion criteria, and random group composition), brain regions investigated, in 1H‐MRS sequences (e.g., STEAM, PRESS, and MEGA‐PRESS) or the scanner (1.5 Tesla, 3 Tesla) (Ajram et al, 2019).…”

Probabilistic Learning of Cue-Outcome Associations is not Influenced by Autistic Traits

“…Though most of the Bayesian and predictive coding models of autism were initially used to describe low-level sensory perception among autistic individuals, they may be applicable to high-level cognitive processes in autism such as learning of statistical regularities and associations as well as prediction and decision-making. Some studies reported that relative to neurotypical individuals, autistic individuals showed differential neural responses to statistical learning of a continuous auditory stream (Scott-Van Zeeland et al, 2010a , 2010b ; Wagley et al, 2020 ) and poorer learning of cue-outcome associations that are needed to make accurate predictions across various paradigms (Amoruso et al, 2019 ; Fogelson et al, 2019 ; Greene et al, 2019 ; Lawson et al, 2017 ; Sapey-Triomphe et al, 2021a , 2021b ), pointing to difficulties autistic individuals may have with extracting regularities from the input and having higher precision to prediction errors. Those models have also been used to understand whether learning differs when the environment changes, which is typically assessed using a probabilistic reversal learning task: learners first learn the contingencies of two cues and their outcomes over many trials (e.g., Cue A is associated with a reward at an 80:20 reinforcement schedule) after which the contingencies change (e.g., now Cue B rewards at an 80:20 reinforcement schedule).…”