Abstract
Scene meaning can be processed extremely quickly, with ‘gist’ extracted even when presentation duration spans only a few dozen milliseconds. This has led some researchers to suggest a primacy of bottom-up information (e.g., Potter et al., AP&P, 2014). However, gist research paradigms have often relied on showing a succession of single, unrelated scene images. This contrasts with our everyday experience in which the world gradually unfolds around us in a predictable manner. Thus, we investigated whether top-down information – in the form of observers’ predictions of an upcoming scene image – facilitates gist processing. If so, predictable scenes should be categorised more readily. Alternatively, if gist primarily relies on bottom-up information then predictability should have a minimal effect. We provided participants (N=129) with five sequential scene images, organised to represent a journey through an environment (e.g., walking down a sidewalk). A final target scene was then presented, either congruent or incongruent with the destination of the journey (e.g., a store interior or a bedroom), followed by a 6AFC response screen. Target scenes were presented for a limited duration (35-250ms), allowing us to delineate the influence of predictions on scene processing over time. As hypothesised, congruent trials were associated with significantly higher performance, especially at shorter durations. We then investigated the neural signature of predictability on scene processing using ERP (N=26), with the same paradigm except with target scenes shown for 1s. Differences were found in the scene-selective ERP marker related to integrating visual properties (P2 component), as well as later components related to contextual regularities including semantic and syntactic meaning (N400 and P600, respectively). Taken together, these results strongly suggest that in real-world situations, top-down predictions of an upcoming scene influence even the earliest stages of its processing, affecting both the integration of visual properties and meaning.