Abstract
Research has shown that event models, a working memory mental representation of actions, are constructed and can be used to predict future actions (Zacks, Kurby, Eisenberg, & Haroutunian, 2011). Although event model construction has been hypothesized, we still know little about its time-course. Namely, how long does it take to construct an event model? Previous research has indicated scene and action categorization may form the basis of an event model (Larson, Hendry, & Loshcky, 2013). If so, then recognizing the scene category (e.g., Park) would prime scene categorization for a subsequently presented scene image. Conversely, some research has shown that top-down knowledge does not aid scene categorization (Potter, Wyble, Hagmann, & McCourt, 2013). If so, then priming would not be observed. Four scene categories were used (Park, Yard, Office, and Kitchen). Participants were presented with 192 trials. A trial presented a variable duration prime image (106, 188, 353, 1024, 2025, or 4026 ms) followed by a target scene image for 47 ms and then a cue word. Participants were instructed to determine if the cue word matched the target. Half of the trials were validly cued, requiring a yes response, whereas the remaining were invalidly cued, requiring a no response. Half the trials presented prime and target images from the same scene category, whereas the remaining trials presented prime and targets from different scene categories. Results show evidence of positive priming when the prime was presented for 188 ms. However, negative priming was found for primes presented for 2025 ms. This suggests that event model construction occurs rapidly and aids scene categorization for subsequent scene images. However, this benefit is limited to a time-course representative of a single eye fixation.
Meeting abstract presented at VSS 2017