Deep Neural Network - Attentional Cueing EEG

This Modern EEG experiment is an expansion on previous work at UvA. Using a Deep Neural Network to examine if controlling for Attentional Cueing before a stimuli will improve the model. Eye-tracking and EEG was used.
I lead a team in designing the paradigm, I built the entirety of the experiment using EventIDE, and managed the lab with two student interns for 60 participants.
DESIGN
The design is quite complex. A fixation cross is presented along with an eye-gaze area for participants to maintain. If gaze is maintained for 250ms, an attentional cue in the form of an arrow will be presented for 150ms followed by the stimuli. If the eye gaze drifted out of the fixation area during the cue, the experiment would reset to the fixation event. The stimuli is an image of modern Amsterdam. The participants are tasked with identifying if a target category is present (bicycles, benches, or vans). The stimuli is only shown for 150ms to prevent direct attention of the target image and thus force participants to use memory recall to identify if the target was present in the image. This was to bolster top-down cognition in which attentional cueing will have an effect. If gaze fixation broke during stimuli presentation the trial was rejected during analysis. After stimulus presentation the participant responded by pressing a button if the target was there.
The complexity of the experiment comes into play when controlling which image was shown. 150 images are split into three categories and shown with 8 different conditions making them unique. Half of these conditions were repeats and half weren't. While maintaining random ordering for every participant, the randomizing of images and conditions became very complex with some unique code. There were also catch trials and a Passive Block. In total, 5280 images were shown and the experiment took two sessions of 3.5 hours.
RESULTS
EEG data represented in an ERP is currently in development.