Humans effortlessly make quick and accurate perceptual decisions about the nature of their immediate visual environment, such as the category of the scene they face. Previous research has revealed a rich set of cortical representations potentially underlying this feat. However, it remains unknown which of these representations are suitably formatted for decision-making. Here, we approached this question empirically and computationally, using neuroimaging and computational modelling. For the empirical part, we collected electroencephalography (EEG) data and reaction times from human participants during a scene categorization task (natural vs. man-made). We then related neural representations to behaviour using a multivariate extension of signal detection theory. We observed a correlation specifically between ~100 ms and ~200 ms after stimulus onset, suggesting that the neural scene representations in this time period are suitably formatted for decision-making. For the computational part, we evaluated a recurrent convolutional neural network (RCNN) as a model of brain and behaviour. Unifying our previous observations in an image-computable model, the RCNN predicted well the neural representations, the behavioural scene categorization data, as well as the relationship between them. Our results identify and computationally characterize the neural and behavioural correlates of scene categorization in humans.

Biological systems must allocate limited perceptual resources to relevant elements in their environment. In vision, this can be achieved by attending to a spatial location or a defining feature (e.g. colour). Spatial attention can be conceived as a spotlight in 2D space, whose focus can differ in breadth and concentration. Could attention to colour also function as a spotlight that highlights an area of colour space? We examined this in an experiment that used multicoloured displays with four overlapping random dot kinematograms (RDKs) that differed only in hue. We manipulated (1) requirement to focus attention to a single colour (red or green) or divide it between two colours (red and green); (2) distances of distractor hues from target hues. This resulted in three different colour contexts, which differed in proximity between targets and distractors. We conducted a behavioural and an electroencephalographic experiment, in which each colour was tagged by a specific flicker frequency and driving its own steady-state visual evoked potential. Behavioural and neural indices of attention showed several major consistencies. Concurrent selection of red and green reduced the efficiency of target enhancement and produced context-dependent distractor interference, consistent with a combination of two hue-selective spotlights of unchanged breadth but reduced concentration. This conceptual similarity in implementation of spatial and feature-based selection implies that attentional modulation of signals in biological systems is likely to be implemented using the same domain-general selective mechanisms, which are well-described with a spotlight metaphor.

No abstract

Symmetry perception studies have generally used two stimulus types: figural and dot patterns. Here, we designed a novel figural stimulus—a wedge pattern—made of centrally aligned pseudorandomly positioned wedges. To study the effect of pattern figurality and colour on symmetry perception, we compared symmetry detection in multicoloured wedge patterns with nonfigural dot patterns in younger and older adults. Symmetry signal was either segregated or nonsegregated by colour, and the symmetry detection task was performed under two conditions: with or without colour-based attention. In the first experiment, we compared performance for colour-symmetric patterns that varied in the number of wedges (24 vs. 36) and number of colours (2 vs. 3) and found that symmetry detection was facilitated by attention to colour when symmetry and noise signals were segregated by colour. In the second experiment, we compared performance for wedge and dot patterns on a sample of younger and older participants. Effects of attention to colour in segregated stimuli were magnified for wedge compared with dot patterns, with older and younger adults showing different effects of attention to colour on performance. Older adults significantly underperformed on uncued wedge patterns compared with dot patterns, but their performance improved greatly through colour cueing, reaching performance levels similar to young participants. Thus, while confirming the age-related decline in symmetry detection, we found that this deficit could be alleviated in figural multicoloured patterns by attending to the colour that carries the symmetry signal.