How learning might strengthen existing visual object representations in human object-selective cortex

How Visual Expertise Changes Representational Geometry: A Behavioral and Neural Perspective

Many people develop expertise in specific domains of interest, such as chess, microbiology, radiology, and, the case in point in our study: ornithology. It is poorly understood to what extent such expertise alters brain function. Previous neuroimaging studies of expertise have typically focused upon the category level, for example, selectivity for birds versus nonbird stimuli. We present a multivariate fMRI study focusing upon the representational similarity among objects of expertise at the subordinate level. We compare the neural representational spaces of experts and novices to behavioral judgments. At the behavioral level, ornithologists (n = 20) have more fine-grained and task-dependent representations of item similarity that are more consistent among experts compared to control participants. At the neural level, the neural patterns of item similarity are more distinct and consistent in experts than in novices, which is in line with the behavioral results. In addition, these neural patterns in experts show stronger correlations with behavior compared to novices. These findings were prominent in frontal regions, and some effects were also found in occipitotemporal regions. This study illustrates the potential of an analysis of representational geometry to understand to what extent expertise changes neural information processing.

Symbolic categorization of novel multisensory stimuli in the human brain

When primates (both human and non-human) learn to categorize simple visual or acoustic stimuli by means of non-verbal matching tasks, two types of changes occur in their brain: early sensory cortices increase the precision with which they encode sensory information, and parietal and lateral prefrontal cortices develop a categorical response to the stimuli. Contrary to non-human animals, however, our species mostly constructs categories using linguistic labels. Moreover, we naturally tend to define categories by means of multiple sensory features of the stimuli. Here we trained adult subjects to parse a novel audiovisual stimulus space into 4 orthogonal categories, by associating each category to a specific symbol. We then used multi-voxel pattern analysis (MVPA) to show that during a cross-format category repetition detection task three neural representational changes were detectable. First, visual and acoustic cortices increased both precision and selectivity to their preferred sensory feature, displaying increased sensory segregation. Second, a frontoparietal network developed a multisensory object-specific response. Third, the right hippocampus and, at least to some extent, the left angular gyrus, developed a shared representational code common to symbols and objects. In particular, the right hippocampus displayed the highest level of abstraction and generalization from a format to the other, and also predicted symbolic categorization performance outside the scanner. Taken together, these results indicate that when humans categorize multisensory objects by means of language the set of changes occurring in the brain only partially overlaps with that described by classical models of non-verbal unisensory categorization in primates.

Expert Tool Users Show Increased Differentiation between Visual Representations of Hands and Tools

The idea that when we use a tool we incorporate it into the neural representation of our body (embodiment) has been a major inspiration for philosophy, science, and engineering. While theoretically appealing, there is little direct evidence for tool embodiment at the neural level. Using functional magnetic resonance imaging (fMRI) in male and female human subjects, we investigated whether expert tool users (London litter pickers: n = 7) represent their expert tool more like a hand (neural embodiment) or less like a hand (neural differentiation), as compared with a group of tool novices (n = 12). During fMRI scans, participants viewed first-person videos depicting grasps performed by either a hand, litter picker, or a non-expert grasping tool. Using representational similarity analysis (RSA), differences in the representational structure of hands and tools were measured within occipitotemporal cortex (OTC). Contrary to the neural embodiment theory, we find that the experts group represent their own tool less like a hand (not more) relative to novices. Using a case-study approach, we further replicated this effect, independently, in five of the seven individual expert litter pickers, as compared with the novices. An exploratory analysis in left parietal cortex, a region implicated in visuomotor representations of hands and tools, also indicated that experts do not visually represent their tool more similar to hands, compared with novices. Together, our findings suggest that extensive tool use leads to an increased neural differentiation between visual representations of hands and tools. This evidence provides an important alternative framework to the prominent tool embodiment theory.SIGNIFICANCE STATEMENT It is commonly thought that tool use leads to the assimilation of the tool into the neural representation of the body, a process referred to as embodiment. Here, we demonstrate that expert tool users (London litter pickers) neurally represent their own tool less like a hand (not more), compared with novices. Our findings advance our current understanding for how experience shapes functional organization in high-order visual cortex. Further, this evidence provides an alternative framework to the prominent tool embodiment theory, suggesting instead that experience with tools leads to more distinct, separable hand and tool representations.

Object recognition is enabled by an experience-dependent appraisal of visual features in the brain's value system

This paper addresses perceptual synthesis by comparing responses evoked by visual stimuli before and after they are recognized, depending on prior exposure. Using magnetoencephalography, we analyzed distributed patterns of neuronal activity - evoked by Mooney figures - before and after they were recognized as meaningful objects. Recognition induced changes were first seen at 100-120 ​ms, for both faces and tools. These early effects - in right inferior and middle occipital regions - were characterized by an increase in power in the absence of any changes in spatial patterns of activity. Within a later 210-230 ​ms window, a quite different type of recognition effect appeared. Regions of the brain's value system (insula, entorhinal cortex and cingulate of the right hemisphere for faces and right orbitofrontal cortex for tools) evinced a reorganization of their neuronal activity without an overall power increase in the region. Finally, we found that during the perception of disambiguated face stimuli, a face-specific response in the right fusiform gyrus emerged at 240-290 ​ms, with a much greater latency than the well-known N170m component, and, crucially, followed the recognition effect in the value system regions. These results can clarify one of the most intriguing issues of perceptual synthesis, namely, how a limited set of high-level predictions, which is required to reduce the uncertainty when resolving the ill-posed inverse problem of perception, can be available before category-specific processing in visual cortex. We suggest that a subset of local spatial features serves as partial cues for a fast re-activation of object-specific appraisal by the value system. The ensuing top-down feedback from value system to visual cortex, in particular, the fusiform gyrus enables high levels of processing to form category-specific predictions. This descending influence of the value system was more prominent for faces than for tools, the fact that reflects different dependence of these categories on value-related information.

Adaptation aftereffects reveal how categorization training changes the encoding of face identity

Previous research suggests that learning to categorize faces along a novel dimension changes the perceptual representation of such dimension, increasing its discriminability, its invariance, and the information used to identify faces varying along the dimension. A common interpretation of these results is that categorization training promotes the creation of novel dimensions, rather than simply the enhancement of already existing representations. Here, we trained a group of participants to categorize faces that varied along two morphing dimensions, one of them relevant to the categorization task and the other irrelevant to the task. An untrained group did not receive such categorization training. In three experiments, we used face adaptation aftereffects to explore how categorization training changes the encoding of face identities at the extremes of the category-relevant dimension and whether such training produces encoding of the category-relevant dimension as a preferred direction in face space. The pattern of results suggests that categorization training enhances the already existing norm-based coding of face identity, rather than creating novel category-relevant representations. We formalized this conclusion in a model that explains the most important results in our experiments and serves as a working hypothesis for future work in this area.

An investigation of the effect of temporal contiguity training on size-tolerant representations in object-selective cortex

The human visual system has a remarkable ability to reliably identify objects across variations in appearance, such as variations in viewpoint, lighting and size. Here we used fMRI in humans to test whether temporal contiguity training with natural and altered image dynamics can respectively build and break neural size tolerance for objects. Participants (N ​= ​23) were presented with sequences of images of "growing" and "shrinking" objects. In half of the trials, the object also changed identity when the size change happened. According to the temporal contiguity hypothesis, and studies with a similar paradigm in monkeys, this training process should alter size tolerance. After the training phase, BOLD responses to each of the object images were measured in the scanner. Neural patterns in LOC and V1 contained information on size, similarity and identity. In LOC, the representation of object identity was partially invariant to changes in size. However, temporal contiguity training did not affect size tolerance in LOC. Size tolerance in human object-selective cortex is more robust to variations in input statistics than expected based on prior work in monkeys supporting the temporal contiguity hypothesis.

An investigation of far and near transfer in a gamified visual learning paradigm

After training, visual perceptual learning improvements are mostly constrained to the trained stimulus feature and retinal location. The aim of this study is to construct an integrated paradigm where the visual learning happens in a more natural context and in parallel for multiple stimulus types, and to test the generalization of learning-related improvements towards untrained features, locations, and more general cognitive domains. Half the subjects were trained with a gamified perceptual learning paradigm for ten hours, which consisted of an orientation discrimination task and a novel object categorization task embedded in a three-dimensional maze. A second group of subjects, an active control group, played ten hours of Candy Crush Saga. Before and after training, all subjects completed a 'near transfer' orientation discrimination and novel object categorization task, as well as a set of 'far transfer' general cognitive and attentional tasks. During the perceptual learning tasks, two different stimulus features and two retinal location pairs were assessed in each task. For the experimental group, one stimulus feature and retinal location pair was trained, whilst the other one remained untrained. Both features and location pairs were untrained in the control group. Far transfer did occur in some domains across all subjects irrespective of the training regimen (i.e. executive functioning, mental rotation performance, and multitask performance and speed). Near transfer was present in both groups, however only more pronounced for one particular task in the experimental group, namely novel object categorization. To conclude, all but one near transfer task did not generalize more than the control group.

Factors Determining Where Category-Selective Areas Emerge in Visual Cortex

A hallmark of functional localization in the human brain is the presence of areas in visual cortex specialized for representing particular categories such as faces and words. Why do these areas appear where they do during development? Recent findings highlight several general factors to consider when answering this question. Experience-driven category selectivity arises in regions that have: (i) pre-existing selectivity for properties of the stimulus, (ii) are appropriately placed in the computational hierarchy of the visual system, and (iii) exhibit domain-specific patterns of connectivity to nonvisual regions. In other words, cortical location of category selectivity is constrained by what category will be represented, how it will be represented, and why the representation will be used.

Neural activity in human visual cortex is transformed by learning real world size

The way that our brain processes visual information is directly affected by our experience. Repeated exposure to a visual stimulus triggers experience-dependent plasticity in the visual cortex of many species. Humans also have the unique ability to acquire visual knowledge through instruction. We introduced human participants to the real-world size of previously unfamiliar species, and to the functional motion of novel tools, during a functional magnetic resonance imaging scan. Using machine learning, we compared activity patterns evoked by images of the new items, before and after participants learned the animals' real-world size or tools' motion. We found that, after acquiring size information, participants' visual activity patterns for the new animals became more confusable with activity patterns evoked by similar-sized known animals in early visual cortex, but not in ventral temporal cortex, reflecting an influence of new size knowledge on posterior, but not anterior, components of the ventral stream. In contrast, learning the functional motion of new tools did not lead to an equivalent change in recorded activity. Finally, the time-points marked by evidence of new size information in early visual cortex were more likely to show size information and greater activation in the right angular gyrus, a key hub of semantic knowledge and spatial cognition. Overall, these findings suggest that learning an item's real-world size by instruction influences subsequent activity in visual cortex and in a region that is central to semantic and spatial brain systems.

Domain-general and domain-specific neural changes underlying visual expertise

Visual expertise induces changes in neural processing for many different domains of expertise. However, it is unclear how expertise effects for different domains of expertise are related. In the present fMRI study, we combine large-scale univariate and multi-voxel analyses to contrast the expertise-related neural changes associated with two different domains of expertise, bird expertise (ornithology) and mineral expertise (mineralogy). Results indicated distributed expertise-related neural changes, with effects for both domains of expertise in high-level visual cortex and effects for bird expertise even extending to low-level visual regions and the frontal lobe. Importantly, a multivariate generalization analysis showed that effects in high-level visual cortex were specific to the domain of expertise. In contrast, the neural changes in the frontal lobe relating to expertise showed significant generalization, signaling the presence of domain-independent expertise effects. In conclusion, expertise is related to a combination of domain-specific and domain-general changes in neural processing.

Temporal Contiguity Training Influences Behavioral and Neural Measures of Viewpoint Tolerance

Humans can often recognize faces across viewpoints despite the large changes in low-level image properties a shift in viewpoint introduces. We present a behavioral and an fMRI adaptation experiment to investigate whether this viewpoint tolerance is reflected in the neural visual system and whether it can be manipulated through training. Participants saw training sequences of face images creating the appearance of a rotating head. Half of the sequences showed faces undergoing veridical changes in appearance across the rotation (non-morph condition). The other half were non-veridical: during rotation, the face simultaneously morphed into another face. This procedure should successfully associate frontal face views with side views of the same or a different identity, and, according to the temporal contiguity hypothesis, thus enhance viewpoint tolerance in the non-morph condition and/or break tolerance in the morph condition. Performance on the same/different task in the behavioral experiment (N = 20) was affected by training. There was a significant interaction between training (associated/not associated) and identity (same/different), mostly reflecting a higher confusion of different identities when they were associated during training. In the fMRI study (N = 20), fMRI adaptation effects were found for same-viewpoint images of untrained faces, but no adaptation for untrained faces was present across viewpoints. Only trained faces which were not morphed during training elicited a slight adaptation across viewpoints in face-selective regions. However, both in the behavioral and in the neural data the effects were small and weak from a statistical point of view. Overall, we conclude that the findings are not inconsistent with the proposal that temporal contiguity can influence viewpoint tolerance, with more evidence for tolerance when faces are not morphed during training.

Visual Category Learning Results in Rapid Changes in Brain Activation Reflecting Sensitivity to the Category Relation between Perceived Objects and to Decision Correctness

Little is known about the time scales in which sensitivity to novel category identity may become evident in visual and executive cortices in visual category learning (VCL) tasks and the nature of such changes in brain activation. We used fMRI to investigate the processing of category information and trial-by-trial feedback information. In each VCL task, stimuli differed in three feature dimensions. In each trial, either two same-category stimuli or two different-categories stimuli were presented. The participant had to learn which feature dimension was relevant for categorization based on the feedback that followed each categorization decision. We contrasted between same-category stimuli trials and different-category trials and between correct and incorrect categorization decision trials. In each trial, brain activation in the visual stimuli processing phase was modeled separately from activation during the later feedback processing phase. We found activation in the lateral occipital complex, indicating sensitivity to the category relation between stimuli, to be evident in VCL within only few learning trials. Specifically, greater lateral occipital complex activation was evident when same-category stimuli were presented than when different-category stimuli were presented. In the feedback processing phase, greater activation in both executive and visual cortices was evident primarily after “misdetections” of same-category stimuli. Implications regarding the contribution of different learning trials to VCL, and the respective role of key brain regions, at the onset of VCL, are discussed.