Inter-element orientation and distance influence the duration of persistent contour integration

A look into feedback neural computation upon collision selectivity

Physiological studies have shown that a group of locust's lobula giant movement detectors (LGMDs) has a diversity of collision selectivity to approaching objects, relatively darker or brighter than their backgrounds in cluttered environments. Such diversity of collision selectivity can serve locusts to escape from attack by natural enemies, and migrate in swarm free of collision. For computational studies, endeavours have been made to realize the diverse selectivity which, however, is still one of the most challenging tasks especially in complex and dynamic real world scenarios. The existing models are mainly formulated as multi-layered neural networks with merely feed-forward information processing, and do not take into account the effect of re-entrant signals in feedback loop, which is an essential regulatory loop for motion perception, yet never been explored in looming perception. In this paper, we inaugurate feedback neural computation for constructing a new LGMD-based model, named F-LGMD to look into the efficacy upon implementing different collision selectivity. Accordingly, the proposed neural network model features both feed-forward processing and feedback loop. The feedback control propagates output signals of parallel ON/OFF channels back into their starting neurons, thus makes part of the feed-forward neural network, i.e. the ON/OFF channels and the feedback loop form an iterative cycle system. Moreover, the feedback control is instantaneous, which leads to the existence of a fixed point whereby the fixed point theorem is applied to rigorously derive valid range of feedback coefficients. To verify the effectiveness of the proposed method, we conduct systematic experiments covering synthetic and natural collision datasets, and also online robotic tests. The experimental results show that the F-LGMD, with a unified network, can fulfil the diverse collision selectivity revealed in physiology, which not only reduces considerably the handcrafted parameters compared to previous studies, but also offers a both efficient and robust scheme for collision perception through feedback neural computation.

Perceptual bias contextualized in visually ambiguous stimuli

The visual appearance of an object is a function of stimulus properties as well as perceptual biases imposed by the observer. The context-specific trade-off between both can be measured accurately in a perceptual judgment task, involving grouping by proximity in ambiguous dot lattices. Such grouping depends lawfully on a stimulus parameter of the dot lattices known as their aspect ratio (AR), whose effect is modulated by a perceptual bias representing the preference for a cardinal orientation. In two experiments, we investigated how preceding context can lead to bias modulation, either in a top-down fashion via visual working memory (VWM) or bottom-up via sensory priming. In Experiment 1, we embedded the perceptual judgment task in a change detection paradigm and studied how the factors of VWM load (complexity of the memory array) and content (congruency in orientation to the ensuing dot lattice) affect the prominence of perceptual bias. A robust vertical orientation bias was observed, which was increased by VWM load and modulated by congruent VWM content. In Experiment 2, dot lattices were preceded by oriented primes. Here, primes regardless of orientation elicited a vertical orientation bias in dot lattices compared to a neutral baseline. Taken together, the two experiments demonstrate that top-down context (VWM load and content) effectively controls orientation bias modulation, while bottom-up context (i.e., priming) merely acts as an undifferentiated trigger to perceptual bias. These findings characterize the temporal context sensitivity of Gestalt perception, shed light on the processes responsible for different perceptual outcomes of ambiguous stimuli, and identify some of the mechanisms controlling perceptual bias.

Perceptual load modulates contour integration in conscious and unconscious states

Previous research has documented that contour detection and integration may either be affected by local features such as the distances between elements or by high-level cognitive factors such as attention in our visual system. Less is known about how low and high level factors interact to influence contour integration. In this paper, we investigated how attention modulates contour integration through saliency (different element spacing) and topological propert ies (circle or S-shaped) when the state of conscious awareness is manipulated. A modified inattentional blindness (IB) combined with the Posner cuing paradigm was adopted in our three-phased experiment (unconscious-training-conscious). Attention was manipulated with high or low perceptual load for a foveal go/no-go task. Cuing effects were utilized to assess the covert processing of contours prior to a peripheral orientation discrimination task. We found that (1) salient circles and S-contours induced different cuing effects under low perceptual load but not with high load; (2) no consistent pattern of cuing effects was found for non-salient contours in all the conditions; (3) a positive cuing effect was observed for salient circles either consciously or unconsciously while a negative cuing effect occurred for salient S-contours only consciously. These results suggest that conscious awareness plays a pivotal role in coordinating a closure effect with the level of perceptual load. Only salient circles can be successfully integrated in an unconscious state under low perceptual load although both salient circles and S-contours can be done consciously. Our findings support a bi-directional mechanism that low-level sensory features interact with high-level cognitive factors in contour integration.