Is Rapid Efficient Scene Perception Also Deep, and Does Attention Help?

Humans perceive natural scenes with great efficiently, if the task is simple (e.g. categorization or gist). Does human efficiency extend to deeper scene perception, involving a wide bandwidth of perceptual experience? We examined this issue while also examining an outstanding issue in attention: Does attention facilitate the encoding of complex perceptual representations during rapid scene perception? Existing research indicates that attention, attracted by an exogenous cue, aids some basic visual processes. But it is not clear that it will aid the encoding of a high-level representation such as that of a complex scene. We adapted the Fei-Fei full report method. Observers viewed a brief scene (in the outer fovea) and wrote what they saw on each trial. We used cartoon-scenes, to encourage rapid scene perception. The scene was presented for 100 ms, preceded by an exogenous attention cue for 75 ms. The cue was a picture frame in either the same location as the picture or the opposite location (invalid cue) in Experiment 1. In Experiment 2, the cue was the same-location frame or no cue at all during the period. The reports were scored by judges blind to condition, and incorrect content was subtracted from correct content. In Experiment 1, the mean corrected score was 8.4 meaningful words with an invalid cue, and 23.3% more correct words with the valid cue (p= .02). In Experiment 2, the mean corrected score was 9.8 meaningful words with no cue, and 13.3% more correct words with the valid cue (p= .03). Thus, in each case a valid attention cue led to better scene perception. This indicates that attention increases the depth of rapid scene perception, producing a more complete understanding of the information in the scene