abstract Tomas Knapen

Tomas Knapen (Vrije Universiteit Amsterdam, the Netherlands)

Fitting event-related response shapes in fMRI and pupil recordings

fMRI has become a dominant modality for cognitive neuroscience over the past 20 years, and pupil dilation recordings have recently increased in popularity as well.  A critical assumption of the majority of fMRI and pupil dilation analyses is that the measured signal is generated by a neural process, which is convolved with a standard impulse response function. Although assuming a ‘canonical’ response shape simplify analyses, it may introduce biases in our results, and sacrifices specificity. Estimation of the shape of event-related responses is possible, but this is seen as complex and the data requirements are often seen as prohibitive. In my talk, I will argue that these event-related responses are not prohibitively complex, and more importantly, are often necessary in fMRI and pupil size experiments. Using several examples from the domains of reward processing and bistable perception, I show that in many cases, the shape of the event-related response differs strongly from the normally assumed canonical response. Thus, the explicit estimation of response shapes is necessary to fully appreciate the underlying neural processes. Based on these findings, I argue that estimation of event-related response shapes is a fundamental step in all signal analyses. Furthermore, I will outline, in tutorial form, a simple set of methods to perform these analysis, implemented in open software.