Advances in visual perceptual learning and plasticity

Abstract

Visual perceptual learning (VPL) is defined as a long-term improvement in performance on a visual task. In recent years, the idea that conscious effort is necessary for VPL to occur has been challenged by research suggesting the involvement of more implicit processing mechanisms, such as reinforcement-driven processing and consolidation. In addition, we have learnt much about the neural substrates of VPL and it has become evident that changes in visual areas and regions beyond the visual cortex can take place during VPL.

Figure 1. Processing during VPL training

According to the model presented, visual perceptual learning (VPL) of the presented visual feature occurs when a bottom-up signal from the feature is boosted by attention (left) or by reinforcement signals (right). Attention enhances task-relevant signals and inhibits task-irrelevant signals, leading to task-relevant VPL. By contrast, reinforcement signals are diffusive and enhance signals from any stimulus feature presented in the visual field, irrespective of whether the feature is task-relevant or task-irrelevant.

Figure 2. Typical tasks used in VPL studies

a | In the Vernier acuity task, a configuration of two or three vertical lines (or dots) is presented. The subject is asked to indicate whether the lines (or the dots) are aligned. b | The RSVP (rapid serial visual presentation) task with moving dots during training is used to examine visual perceptual learning (VPL) of task-irrelevant coherent motion. The subject is asked to identify two target items (such as white letters) within a sequence of non-target items (such as black letters) at the centre of the display. The background display consists of coherent motion (dots moving in the same direction at the same speed) and random motion (dots moving in random directions with random speed). The arrows represent the velocity of the coherent motion. In test stages before and after training, only coherent motion is displayed (not shown here) to determine how performance in motion-discrimination or -detection tasks is changed by training. c | A texture-discrimination task is the most frequently used task in VPL studies. The subject is first asked to respond according to whether a ‘T’ (as shown in the figure) or an ‘L’ is presented in the centre of the display to ensure fixation at the centre, and then to indicate whether the orientation of the target (the three elements with orientation that differs from that of the rest of the elements) is vertical (as shown in the figure) or horizontal. VPL of the target orientation is examined. Part b is modified, with permission, from REF. © (2001) Macmillan Publishers Ltd. All rights reserved. Part c is modified, with permission, from REF. © (1991) National Academy of Sciences.

Figure 3. Neural correlates of VPL

The regions of the brain thought to be altered by visual perceptual learning (VPL). Some experiments have indicated that training on a visual task changes visual representations in the early stages of visual signal processing, such as the tuning properties and activity of the primary visual cortex (V1) region that retinotopically corresponds to the location of the trained stimulus in the visual field. Others have instead suggested that training alters the weight of connections (ω₁, ω₂ … ω_i) between the visual cortex and regions of the brain involved in decision making, or within the decision-making regions themselves. In VPL of motion or a feature carrying spatial information, the weight changes may predominantly occur between areas in the higher visual cortex, such as the middle temporal area (MT) and the lateral intraparietal area (LIP), which is thought to be involved in visual decision-making processes. MT is usually responsible for coarse binocular disparity (depth) processing; however, when MT is inactivated, decision-making regions may learn to give more weight to signals from areas involved in ventral processing, including V4, when discriminating coarse binocular disparity.