Amelia Woods
Affiliation: University of Exeter
Category: Philosophy
Keywords: Autism, Trauma, PTSD, Memory, Perception
Date: Wednesday 3rd of September
Time: 18:30
Location: Room 232 (232)
View the full session: Self-Knowledge
There are various accounts of autism that appeal to predictive processing. In this work, I will categorise them into Stronger Bottom-Up (SBU) (see Lawson, Rees and Friston, 2014; Van de Cruys et al., 2014) and Weaker Top-Down (WTD) (see Pellicano and Burr, 2012) accounts. It is generally unclear as to how we should decide what the more accurate view is. Both WTD and SBU accounts propose an imbalance between higher and lower levels of the hierarchy, meaning that the influence of priors on perception is atypical. In general, it is assumed that WTD and SBU accounts are functionally equivalent (Lawson, Rees and Friston, 2014, p. 6; Angeletos Chrysaitis and Seriès, 2023). In this paper, I examine the tenfold higher risk of PTSD in autistic people (Rumball, Happé and Grey, 2020; Rumball et al., 2021) and, drawing on predictive processing accounts of PTSD, argue that SBU accounts of autism can better accommodate this epidemiological pattern. All other things being equal, this provides a way of adjudicating between SBU and WTD accounts in favour of the former.
The fact that autistic people as a group experience higher rates of various adverse life experiences and events considered to be traumatic (Rumball et al., 2021; Douglas and Sedgewick, 2024) is not enough to explain the tenfold increase in the occurrence of PTSD relative to the neurotypical population. An explanation of what it is about autistic cognition that makes it more susceptible to PTSD is needed. This piece of work will give a detailed discussion of how SBU accounts of autism can explain factors hypothesised to make autistic people more vulnerable to PTSD. The development and persistence of PTSD is thought to involve a consistent failure of learning and contextualisation during various processes. The processes I will discuss are 1) processing (internal and external) sensory input 2) mental representation (modelling the self, the world, and self-in-world) and 3) memory encoding and recall.
First, I will argue that, for PTSD to present, a hypothesis relating to the traumatic event must be formed, given unusually high prior probability and inform perception. Whether we use SBU or WTD accounts, autism involves increased precision on bottom-up sensory signals. In WTD accounts, this is due to the failure of priors to constrain perception. In SBU accounts, this is because high inflexible precision on prediction error results in overestimating the number of changes in worldly regularities. Increased attention to sensory input at the time of the event means that more sensory objects are available to act as a trigger. In other words, the prior related to the traumatic event is overly sensory-informed. However, SBU accounts suggest a normal (or amplified) capacity to form priors paired with increased bottom-up processing. This means that SBU accounts give a better explanation of autistic people’s vulnerability to PTSD.
Then, I will discuss how representations of the self in the present are constructed by the brain. Crucially, I will highlight the consequences of certain representations of the self on action selection and vice versa. The consequences of delayed action selection and negative valuations of one’s actions are important for understanding the persistence of PTSD. I will look at tonic immobility (TI), avoidance behaviours, and reenacting behaviours specifically. I will argue that TI at the time of a threatening event is made more likely by autism. I will also argue that re-enacting and avoidance behaviours become more likely in autism because when the meta-learning necessary to ignore prediction error isn’t present, the best way to minimise prediction error is to “harvest sensory signals that we can predict” (Friston, Thornton and Clark, 2012), avoiding ones that we can’t.
Finally, I will give a predictive processing explanation of memory differences in autism, highlighting differences during encoding, storage and retrieval, as well as at different levels of the memory hierarchy (perceptual, semantic, episodic). My account of memory will be informed by phenomenological accounts from autistic people as well as empirical studies. Ultimately, it aligns with the cognitive model of PTSD which states that people who engage in primarily data-driven processing during The Event are more likely to develop persistent PTSD (Ehlers and Clark, 2000). It also has the benefit of explaining the following three observations about autistic memory: 1) differentiating between generally vivid autobiographical memories and intrusive thoughts or flashbacks is hard for autistic adults (Rumball et al., 2024, p. 178), 2) autistic people have difficultly forming and recollecting episodic memories (Griffin, Bauer and Gavett, 2022; Norris and Maras, 2022), and, 3) Autistic people have normal to enhanced perceptual memory and pattern recognition abilities but more effortful/diminished intentional recall in comparison to neurotypical people (Desaunay et al., 2020).