Current models suggest that perception is a decision process: given noisy perceptual signals, the brain has to decide what they represent. While attention is known to enhance the perceptual signal, it has been unclear how it modulates the decision mechanism itself. Here we explored this issue in a series of studies. We used a spatial cuing paradigm to manipulate the attentional focus of observers, and found that attention leads to a conservative detection criterion such that attended stimuli are reported less often than unattended ones (Chapter 1). We investigated whether this effect would generalize to situations that do not involve detection tasks by using the same cuing paradigm, but instead asking observers to discriminate between two stimulus categories. We found that attention leads to low subjective ratings of visibility (Chapter 2). In both sets of experiments, the results were strongest when detection or discrimination capacity d' was equated between different levels of attention, or when stimuli had low contrast. To account for these results, we developed a variance reduction (VR) model of attention in which attention is postulated to reduce the variability of the perceptual signal, while keeping the decision criteria constant (Chapter 3). The VR model provided a good fit to the data observed in Chapters 1 and 2. We tested critical assumptions of the model using functional magnetic resonance imaging (Chapter 4). We found that high activity in the dorsal attention network (DAN) in the brain, which is indicative of a high attentional state, led to lower variability in the evoked signal in motion sensitive area MT+, thus supporting the idea that attention reduces perceptual variability. Further, high DAN activity resulted in lower confidence ratings, which confirmed that the findings from Chapter 2 generalize to exogenous attentional fluctuations and are not limited to spatial cuing. We tested the VR model further by extending it beyond the realm of attention (Chapter 5). We used transcranial magnetic stimulation (TMS) to directly increase the variability of the perceptual signal. The effects mirrored the effect of lack of attention: TMS led to decreased performance but increased subjective ratings. Finally, we explored the influence of attention on the amount of information carried by one's subjective ratings. We found that attention made subjective ratings more predictive of accuracy (i.e., attention improved metacognitive sensitivity) despite the fact that it decreased the overall magnitude of the subjective ratings (Chapter 6). To account for this finding, we developed a simple extension to the VR model - the "variance and criterion jitter reduction" (VCJR) model of attention which postulates that attention reduces the amount of trial-to-trial criterion jitter. Computational modeling shows that this reduction of criterion jitter leads to improved metacognitive sensitivity. We discuss these findings in relation to current debates related to attention and subjective perception, and speculate how they may account for our impression that we clearly see everything in our visual fields, including unattended objects that receive little processing.
| Identifer | oai:union.ndltd.org:columbia.edu/oai:academiccommons.columbia.edu:10.7916/D82N50BK |
| Date | January 2012 |
| Creators | Rahnev, Dobromir Asenov |
| Source Sets | Columbia University |
| Language | English |
| Detected Language | English |
| Type | Theses |
Page generated in 0.0019 seconds