V-STIR · Visual-spatiotemporal integration for recognition

Visual-spatiotemporal integration for recognition

Successful interactions entail detecting object features from noisy environment, integrating local features into global forms and identifying them based on previous experience. Despite decades of intense investigations, the brain mechanisms that mediate the translation of sensory information to subjective perception remain largely unknown. Our research has focused on the neural processes that mediate feature integration, object categorisation and perceptual decisions and brain plasticity on these cognitive functions. In particular, the projects of Kuai supervised by Kourtzi aimed to characterise how the brain integrates discrete elements across space and time to form internal templates of objects that support visual recognition. In the first project, Kuai focused on the neural mechanism of spatiotemporal integration under slit-viewing. To successfully achieve slit-viewing object perception, the visual system needs to integrate not only visible information but also fragments viewed and stored in memory. It is intriguing how our brain meets this challenge and pieces together the relevant information across space and time. By combining psychophysics and fMRI, our study found that contour integration was successful even though only a fraction of a single contour element was seen at any given moment, suggesting reliable spatiotemporal integration in the human brain. Similar to spatial contour integration, temporal contour integration obeyed the Gestalt law of good continuation. Furthermore, fMRI experiments showed that temporal contour integration engaged higher dorsal visual areas, intraparietal sulcus (IPS) and lateral occipital cortex (LO), but not early visual cortex. In particular, posterior parietal cortex, played an important role in maintaining image fragments in visual memory in temporal contour integration. These findings provide the first neuroimaging evidence for the brain mechanisms that medaite slit-viewing perception. Comparing these results to previous studies of spatial integration, suggest a unified model for the perceptual integration spatial and temporal information. In the second his project, Kuai aimed to use classification images to probe how learning optimises decision templates in the ventral visual cortex. Classification images have been widely used in behavioural studies; however, application of this methodology to neuroimaging has been limited due to noisy fMRI signals and the small number of samples that can be acquired during fMRI scans. To overcome these limitations, this study developed a new method that uses reverse correlation combined with multivoxel pattern analysis. The method is able to identify the critical image parts that determine neural coding in human brain areas. The results demonstrate that learning tunes discriminative image parts between categories in higher ventral cortex. This work has been submitted to a high-ranked peer-reviewed journal. Further, Kuai conducted a study comparing the ability of young and older adults to improve through training in visual shape discrimination. Perceptual learning has been considered as a potential treatment to prevent age-related decline in visual functions. It has been reported that the ability to learn decreases with age. However, the factors affecting learning in older adults remain unclear. The study found an interesting dissociation in the ability to learn visual forms in aging. In particular, learning improves global form discrimination in both young and older adults. In contrast, learning to integrate local elements is impaired in older age. These findings suggest that visual selection processes rather than global feature representations, provide a fundamental limit for learning-dependent plasticity in the aging brain. The work has been accepted in one of the top peer-reviewed journals in psychology, Psychological Science and was praised by an expert reviewer as 'a most welcome contribution to the ageing literature'. Finally, Kuai has been involved in another three collabourative projects. The first collabourative project investigated the time-course of visual shape learning during hard and easy training and correlates human learning performance with the level of GABA signals in their brains. The study found a distinct time-course for hard compared to easy training and correlations of GABA concentration in higher occipitotmeporal brain regions with task performance. The second project examined the effect of cognitive abilities on human shape learning. In particular, the study correlated human learning ability and performance in cognitive tasks (i.e. memory, attention) to investigate individual variability in learning ability. The results demonstrate that attentional and working memory abilities predict learning performance. The third project investigated the neural mechanisms involved in the processing of shape curvature in the human brain. Findings from these studies advance our understanding of the neural mechanisms that support our ability to utilise previous experience for successful visual recognition. As such, this work has potential implications for translational interventions and training programmes for rehabilitation in healthy ageing and disease.

Текст от CORDIS, на английски · Данни: CORDIS, © Европейски съюз