Evidence-Accumulation Models: Caution and Prior Probability

Spatial attention and prior probability manipulations have been shown to induce response biases in perceptual decision making behavior. Here, we study the interplay between spatial attention and prior probability manipulations. Participants completed a novel two-alternative forced choice task which simultaneously manipulates spatial attention and prior probability in a factorial design. The task features two cues which prompt the participant to attend to one side of the screen and to expect a given stimulus. A preliminary behavioral analysis showed strong influence of spatial attention and weak influence of prior probability on both RT and accuracy. We fit several variants of the diffusion decision model (DDM) to test which cognitive processes are affected by each manipulation. Preliminary results suggests that the prior probability manipulation affects the starting point parameter, and the attention manipulation the drift rate and non-decision times, in line with earlier literature. The results of this study will set the stage for an fMRI project investigating the neural underpinnings of spatial attention and perceptual expectation.

This is an in-person presentation on July 19, 2023 (15:20 ~ 15:40 UTC).

The evidence accumulation models (EAMs) are useful to study cognitive processes and their effects on response times and accuracy, capturing dependencies between the two. One phenomenon of interest is the speed-accuracy trade-off, where individuals sacrifice one for the other. Classical EAMs assume a continuous trade-off between speed and accuracy, thereby allowing performance to vary between guessing and (in principle) almost perfect responding. However, alternative tradition of thinking suggests that participants may switch between distinct states rather than control the trade-off on a continuum. Hidden Markov Models (HMMs) are typically used to describe such behaviour, assuming two states - random guessing and stimulus-controlled states. Typical HMM applications assume that speed and accuracy are independent of each other, conditioned on the states. However, evidence accumulation presumably takes place at the least under the controlled state, inducing a speed-accuracy trade-off within that state. In this talk, we introduce a model that combines a HMM with an EAM that contains a discontinuous speed-accuracy trade-off on a larger scale (between states) and a continuous speed-accuracy trade-off on a smaller scale (within states), and show some applications on empirical data. We’ll also discuss our experiences with a robust Bayesian workflow employed to validate the implementation of the model, and potential extensions to the model and its applications.

This is an in-person presentation on July 19, 2023 (15:40 ~ 16:00 UTC).

The diffusion model has become a standard model for perceptual decision making over the last decades. A challenge for cognitive models for this type of task is to model differences in mean reaction times for correct responses and error responses. In particular, for simple tasks with short response times, incorrect responses typically have lower mean response times than correct responses. In the diffusion model framework, this asymmetry is typically explained by variability in the starting point of the evidence accumulation process. Recently, the Levy-Flight model was introduced as an alternative explanation for fast errors based on jumps in evidence accumulation. In this talk, the goodness-of-fit of the Diffusion Model and the Levy-Flight Model is compared for different tasks.

This is an in-person presentation on July 19, 2023 (16:00 ~ 16:20 UTC).

In typical response time tasks, catch trials are trials in which no stimulus is shown and participants accordingly do not have to respond. In previous studies, it has been assumed that stimulus expectancy–operationalized via the frequency of catch trials–affects response caution. For conditions with a higher proportion of catch trials enhanced response caution is expected. However, higher proportions of catch trials might also lead to less practice regarding the actual binary decision task, manifesting in reduced speed of information processing or longer encoding or motor execution times. By means of diffusion modeling we examine data from one of the studies that aimed at influencing response caution via a catch trial manipulation. Furthermore, we present data from a new study in which we systematically varied the proportion of catch trials. We consistently find longer non-decision times for conditions with higher proportions of catch trials, whereas the pattern is less clear-cut for drift rate and threshold separation. By means of a parameter recovery study, we further show that the effect in non-decision time is not driven by trade-offs in parameter estimation. In sum, the catch trial manipulation might be a questionable manipulation of response caution as it does not selectively influence threshold separation.

This is an in-person presentation on July 19, 2023 (16:20 ~ 16:40 UTC).