Figure 1. Theoretical accounts of decision bias.

(A) Signal-detection-theoretic account of decision bias. Signal and noise + signal distributions are plotted as a function of the strength of internal sensory evidence. The decision point (or criterion) that determines whether to decide signal presence or absence is plotted as a vertical criterion line c, reflecting the degree of decision bias. c can be shifted left- or rightwards to denote a more liberal or conservative bias, respectively (green arrow indicates a shift toward more liberal). (B, C) Drift diffusion model (DDM) account of decision bias, in which decisions are modelled in terms of a set of parameters that describe a dynamic process of sensory evidence accumulation toward one of two decision bounds. When sensory input is presented, evidence starts to accumulate (drift) over time after initialization at the starting point z. A decision is made when the accumulated evidence either crosses decision boundary a (signal presence) or decision boundary 0 (no signal). After a boundary is reached, the corresponding decision can be either actively reported by a button press (e.g. for signal-present decisions), or remain implicit, without a response (for signal-absent decisions). The DDM can capture decision bias through a shift of the starting point of the evidence accumulation process (panel B) or through a shift in bias in the rate of evidence accumulation toward the different choices (panel C). These mechanisms are dissociable through their differential effect on the shape of the reaction time (RT) distributions, as indicated by the curves above and below the graphs for target-present and target-absent decisions, respectively. Panels B. and C. are modified and reproduced with permission from Urai et al., 2018 (Figure 1, published under a CC BY 4.0 license).