Attending to visual information for perception and recognition
A fundamental function of the visual system, and other sensory systems, due to resource limitations, is to optimally select task-relevant information from the barrage of information impinging the retina. In a series of experiments different aspects of attention to visual information for perception and recognition are addressed. Firstly, what information is selected? In Chapter 2, I begin with addressing the generic problem of knowing which information drives the perception of a stimulus. As a case study example, a first experiment using Bubbles (Gosselin and Schyns, 2001) determines the specific information underlying the perception of the stable percepts of an ambiguous image and shows that this information is grounded in different spatial filters processing each image interpretation. A further experiment employs frequency-specific adaptation to induce a perception of the image that is orthogonal to the adapting frequencies, validating this information drives the selective perception of the ambiguous image. Secondly, if we know the subset of information that is selected for the perception of an input, can the processing of information underlying a percept be selectively suppressed, thereby inducing an alternative percept? In the experiments of Chapter 3, I further apply this spatial frequency adaptation method to test the relevance of local image features for the perception of the 2 ambiguous image, and the recognition of gender in hybrid faces. While the results of the experiment on the perception of the ambiguous image suggest an effective method for testing the role of local stimulus information for perception, the results of the experiment on the recognition of gender in faces showed no effect of adaptation region on perception of gender. Thirdly, how does selective information use evolve with learning? And what are the mechanism(s) that enable this learning? In Chapter 4, I investigated the evolution of information use in the discrimination of unfamiliar faces using a perceptual learning paradigm. I used the method of noise masking to examine the facial regions observers used, over time, for successful discrimination, and to determine the mechanisms underlying performance improvements. The results showed that the efficiency of observers to use the information available increased with learning differentially across different regions of the face. Supplementary to examining attention to information in terms of the information content of the stimulus, another series of experiments investigated the effect of attention on the temporal dynamics of processing visual information, and the locus of this effect within the stream of information processing. Using spatial pre-cueing to manipulate attention and a speed-accuracy trade-off (SAT) method to examine the full time-course of visual processing, I analysed, in addition to the behavioural response and the parameters of the SATfunction, the lateralised readiness potential (LRP)and other components of the event-related brain potentials (ERPs) to determine the locus of any attentional modulations on the speed of processing. The results of three experiments showed that attention can speed up visual information processing, and that the locus of this effect is at later processing stages related to the categorisation of a stimulus.