Applying signal detection theory to evoked response potentials for understanding mechanisms of bias and sensitivity in face detection tasks

The present work explores effects related to stimulus contrast and probability in ERP studies of face perception by applying theory and methodology from psychophysics. Behavioral indices and models of visual processing are elucidated in psychophysics, but a link to EEG measures has not been established. A Signal Detection Theory (SDT) approach was used, which models the effects of contrast and stimulus probability in terms of the mechanisms of sensitivity and bias. In Experiment 1, observers completed a face detection task. Face-present and face-absent stimuli comprised the target and distracter trials, respectively. The contrast of the stimuli and the prior probability of a target varied between blocks. There was also an added noise and a zero noise condition. In Experiment 2, the task was the same. The only differences were that there was only a single, low level of noise, the contrast levels for the stimuli were based on contrast thresholds estimated for individuals, and blocks of trials were included at the beginning and end of the experiment to test for possible adaptation effects. Behavioral and EEG data were recorded and analyzed using standard psychophysiological and psychophysics approaches. Behavioral results were relatively straightforward within the SDT framework, such that d' and beta varied with stimulus contrast and probability in predictable ways. Analysis of the EEG data showed predicted effects for most components, although the effects of contrast and prior probability are much more widespread than previously thought. Principles of SDT were used to guide further analyses of the EEG data. Indices of sensitivity and bias were derived and applied; these results were compared to those from the analysis of behavioral data. The results were considered in the context of linking behavioral and EEG data, with special focus given to modeling and measurement issues.