Knowing when to stop: Aberrant precision and evidence accumulation in schizophrenia

Abstract

Predictive coding and active inference formulations of the dysconnection hypothesis suggest that subjects with schizophrenia (SZ) hold unduly precise prior beliefs to compensate for a failure of sensory attenuation. This implies that SZ subjects should both initiate responses prematurely during evidence-accumulation tasks and fail to inhibit their responses at long stop-signal delays. SZ and healthy control subjects were asked to report the timing of billiards-ball collisions and were occasionally required to withhold their responses. SZ subjects showed larger temporal estimation errors, which were associated with premature responses and decreased response inhibition. To account for these effects, we used hierarchical (Bayesian) drift-diffusion models (HDDM) and model selection procedures to adjudicate among four hypotheses. HDDM revealed that the precision of prior beliefs (i.e., starting point) rather than increased sensory precision (i.e., drift rate) drove premature responses and impaired response inhibition in patients with SZ. From the perspective of active inference, we suggest that premature predictions in SZ are responses that, heuristically, are traded off against accuracy to ensure action execution. On the basis of previous work, we suggest that the right insular cortex might mediate this trade-off.

The drift diffusion model as the choice rule in inter-temporal and risky choice: A case study in medial orbitofrontal cortex lesion patients and controls

“…Second, taking into account the full RT distributions can reveal additional information regarding the dynamics of decision processes [14,15]. This is of potential interest, in particular in the context of maladaptive behaviors in clinical populations [14,[22][23][24][25] but also when the goal is to more fully account for how decisions arise on a neural level [10].…”

The drift diffusion model as the choice rule in inter-temporal and risky choice: a case study in medial orbitofrontal cortex lesion patients and controls

“…Second, taking into account the full reaction time distributions can reveal additional information regarding the dynamics of decision processes 13,14 . This is of potential interest, in particular in the context of maladaptive behaviors in clinical populations 13,[22][23][24][25] but also when the goal is to more fully account for how decisions arise on a neural level 9 .…”

Glutamate and Dysconnection in the Salience Network: Neurochemical, Effective-connectivity, and Computational Evidence in Schizophrenia

“…The dysconnection hypothesis (c.f., disconnection hypothesis, 9) 1 states that the psychopathology of schizophrenia should be studied at three levels of analysis: neurochemical, effective-connectivity (network-connectivity), and computational levels. At the computational level, the dysconnection hypothesis states that a suboptimal (e.g., schizophrenia) Bayesian brain (10) would overly afford confidence or precision (i.e., inverse variance) to its predictions about the external stimuli and would overestimate the reliability of the prediction errors (PE), leading to false inferences (e.g., hallucinations) and cognitive failures (e.g., cognitive control, 11). At the effective-connectivity level, a predictive coding algorithm (12), namely, hierarchical message passing between lower and higher cortical levels would be altered in terms of aberrant backward and forward interregional connectivity strength (13).…”