Simulating lesion-dependent functional recovery mechanisms

Abstract

Functional recovery after brain damage varies widely and depends on many factors, including lesion site and extent. When a neuronal system is damaged, recovery may occur by engaging residual (e.g., perilesional) components. When damage is extensive, recovery depends on the availability of other intact neural structures that can reproduce the same functional output (i.e., degeneracy). A system’s response to damage may occur rapidly, require learning or both. Here, we simulate functional recovery from four different types of lesions, using a generative model of word repetition that comprised a default premorbid system and a less used alternative system. The synthetic lesions (i) completely disengaged the premorbid system, leaving the alternative system intact, (ii) partially damaged both premorbid and alternative systems, and (iii) limited the experience-dependent plasticity of both. The results, across 1000 trials, demonstrate that (i) a complete disconnection of the premorbid system naturally invoked the engagement of the other, (ii) incomplete damage to both systems had a much more devastating long-term effect on model performance and (iii) the effect of reducing learning capacity within each system. These findings contribute to formal frameworks for interpreting the effect of different types of lesions.

Degeneracy in the nervous system: from neuronal excitability to neural coding

“…One example is the partial recovery of language after poststroke aphasia [70] wherein the underlying networks involved in language functional recovery [71] can be identified via simulations. [72,73] These degenerate networks include either regions with less activity during normal language processing or regions that are normally engaged with activities other than language. [70] Therefore, plasticity and dynamical reorganization in network functions are keys to supporting degeneracy.…”

A mixed generative model of auditory word repetition

“…In WORM, we combined two distinct types of models. The higher levels were instantiated as partially observable Markov Decision processes (Figure 1) of the type previously used to simulate word repetition in (Sajid et al, 2020b; Sajid et al, 2020a; Sajid et al, 2021b). Conversely, a continuous model was used at the lower (faster) level: specifically, a generative model for synthesising and recognising speech (Friston et al, 2020b).…”

“…Under this framework, one can generate distinct behaviours using different generative models. For example, active inference has been shown to successfully simulate a wide range of complex behaviours, including word repetition with fully discrete models (Sajid et al, 2020a; Sajid et al, 2020b; Sajid et al, 2021b), dyadic exchanges (Friston and Frith, 2015; Friston et al, 2020a), active listening (Friston et al, 2020b), active vision (Parr et al, 2021) and scene construction (Friston et al, 2017d; Parr and Friston, 2017; Heins et al, 2020).…”

“…Turn-taking Representations Discrete state-space active inference models (Sajid et al, 2020c;Sajid et al, 2020b;Sajid et al, 2020a;Sajid et al, 2021b) Partially observable Markov decision process (POMDP) models were instantiated, and outcomes were simulated using prespecified probability distributions.…”

Degeneracy in the neurological model of auditory speech repetition

“…However, even here, the use of the classic pathway varied across task and subregional configurations. These distinct hierarchies are examples of degenerate functional architectures underwriting word repetition 16, 20, 40 . Participants might repeat words/pseudowords by either engaging pOp or using an alternative pathway involving pSTS.…”