Purchase this article with an account.
Marcus Cappiello, Weiwei Zhang; Enumeration and Reentrant Processes. Journal of Vision 2015;15(12):907. doi: 10.1167/15.12.907.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Extracting summary statistics from a visual scene requires little effort. However, it is unclear how fast this process can occur. The present study tested whether the initial feed-forward sweep in early sensory processing is sufficient for statistic perception by interrupting reentrant processes with Object Substitution Masking (OSM). OSM typically uses a 4-dot mask surrounding the briefly presented target, which either persists after the target offset (delayed-offset) or disappears with the target offset (simultaneous-offset). If the target duration is short enough, the 4-dots in the delayed-offset condition will interrupt the reentrant process. Previous literature has primarily relied on accuracy measurements to investigate masked summary statistics, finding mixed results. To resolve this issue, we adopted the dual-process signal detection (DPSD) framework to separate out two aspects of statistical perception: a discrete high-threshold process of seeing and a continuous feeling of sensing (analytically equivalent to recollection and familiarity in recognition memory, respectively). In Experiment 1, several groups of dots were flashed on the screen for 40ms. Participants first reported whether one group of dots contained more dots than other groups, and then reported their confidence on their choice. These responses were used to construct Receiver Operating Characteristics (ROC) from which seeing and sensing parameters were extracted. Three conditions were included: delayed-offset, simultaneous-offset, and no mask. The delayed-offset condition led to a large drop in the seeing component, with a small decrease in sensing, compared to the simultaneous-offset condition, suggesting that reentry processes were necessary for seeing numerosity. This decrease in seeing was eliminated in Experiment 2 when stimulus duration was extended to 80ms to allow reentry processes even for the delayed-offset condition. Taken together, the present results suggest that the initial feed-forward sweep is sufficient for sensing whereas reentrant processes are necessary for seeing in statistic perception.
Meeting abstract presented at VSS 2015
This PDF is available to Subscribers Only