A computational model of inner speech supporting flexible goal-directed behaviour in Autism

Abstract

Experimental and computational studies propose that inner speech boosts categorisation skills and executive functions, making human behaviour more focused and flexible. In addition, many clinical studies highlight a relationship between poor inner-speech and an executive impairment in autism spectrum condition (ASC), but contrasting findings are reported. Here we directly investigate the latter issue through a previously implemented and validated computational model of the Wisconsin Cards Sorting Tests. In particular, the model was applied to explore potential individual differences in cognitive flexibility and inner speech contribution in autistic and neurotypical participants. Our model predicts that the use of inner-speech could increase along the life-span of neurotypical participants but would be reduced in autistic ones. Although we found more attentional failures (i.e., wrong behavioural rule switches) in autistic children/teenagers and more perseverative behaviours in autistic young/older adults, only autistic children and older adults exhibited a lower performance (i.e., fewer consecutive correct rule switches) than matched control groups. Overall, our results corroborate the idea that the reduced use of inner speech could represent a disadvantage for autistic children and autistic older adults. Moreover, the results suggest that cognitive-behavioural therapies should focus on developing inner speech skills in autistic children as this could provide cognitive support throughout their whole life span.

Flexible Goal-directed Cognition and Inner-Speech in Schizophrenia Spectrum Disorders: from clinical data to computational modeling, and backward.

“…In particular, the next section illustrates a theory-driven neuro-inspired computational model that performs the WCST (thegold-standard to probe flexible cognition). We experimentally corroborated the theory and various versions of the model (Granato et al, 2020;Granato and Baldassarre, 2021;Granato et al, 2022) The working memory component stores the 'priorities' of 239 the task sub-goals (e.g., the sorting rules) and determines 240 the probability with which the rules are selected. Working 241 memory priority values undergo a temporal decay, leading 242 them toward the same baseline.…”

“…namely to find the model parameters that best reproduce the behavioural data of the human groups (for more details, see section 1.2 in Supplementary Materials). In our previous studies (Granato et al, 2020;Granato and Baldassarre, 2021;Granato et al, 2022), we ran thousands of simulations and we computed the Minimum Square Error (MSE;…”

“…Interestingly, as 832 we showed in Granato and Baldassarre ( 2021), an 'irra-833 tional/distracted' behaviour leads both to NPE and PE. 834 Therefore, many PE can be computed due to reasoning 835 failures rather than perseverative responses (we defined 836 these errors 'distraction-related PE' in Granato et al, 2022). 837 Summing up, the PE/NPE balance of RI patients could hide 838 an imbalance toward distraction/reasoning failures, thus 839 supporting more complete interpretations of SSD deficits 840 (Li, 2004;Tyburski et al, 2017;Gebreegziabhere et al, 841 2022).…”

“…Fernyhough and Borghi, 2023). In particular, we fixed an onset frequency and the motivational alteration, In Granato et al (2020Granato et al ( , 2022, we demonstrated that the 1062 four WCST indices we consider (CC, PE, NPE, FMS) sup-1063 port an efficient and complete neuropsychological analysis 1064 of the participants. However, in the future additional be-1065 havioural indices may be considered to extract further par-1066 ticipant traits (e.g., the learning-to-learning index; Heaton 1067 and Staff, 1993).…”

“…However, the data-A PREPRINT -NOVEMBER 14, 2023 driven ML method cannot provide information about the data's latent causes (i.e., neuro-cognitive features). Therefore, we analised the clusters data with our theory-based and experimentally validated computational model of the WCST (Granato et al, 2020;Granato and Baldassarre, 2021;Granato et al, 2022). Overall, our integrated datadriven/model-based approach was capable of disentangling different neurocognitive profiles at the basis of SSD subpopulations.…”