Contributed Talk Sessions | Poster Sessions | All Posters | Search Papers

Poster Session B: Wednesday, August 13, 1:00 – 4:00 pm, de Brug & E‑Hall

How to reliably estimate collapsing-threshold diffusion model parameters? A simulation study

Amir Hosein Hadian Rasanan¹, Lukas Schumacher¹, Michael Nunez², Gabriel Weindel³, Joerg Rieskamp; ¹University of Basel, ²University of Amsterdam, ³Université de Lausanne

Presenter: Amir Hosein Hadian Rasanan

Evidence accumulation models have become the dominant theory in explaining neural and behavioral constructs of decision-making. The main principle of these models is that a decision-maker accumulates noisy evidence until a constant threshold is reached. However, several behavioral and neuroscientific findings, besides some theoretical motivations like optimality, have led to alternative proposals, such as ``collapsing threshold" models. Usually, these models offer a more accurate fit to empirical data. However, a major issue with these models is the unreliability of parameter estimation. Due to this, researchers have relied solely on model fit comparisons, avoiding interpretation of the parameter values -- leading to controversial findings in the literature that support these models. This work introduces a reliable model estimation framework by linking the non-decision time to external measurements. In this modeling framework, we consider a joint likelihood function for behavioral measurements and the non-decision time measurement, constraining the non-decision time estimation. The results of a parameter recovery study showed that the proposed joint model makes the collapsing threshold parameters identifiable.

Topic Area: Reward, Value & Social Decision Making

Extended Abstract: Full Text PDF