Learning and Representation of Hierarchical Concepts in Hippocampus and Prefrontal Cortex.

A key aspect of conceptual knowledge is that it can be flexibly applied at different levels of abstraction, implying a hierarchical organization. It is yet unclear how this hierarchical structure is acquired and represented in the brain. Here we investigate the computations underlying the acquisition and representation of the hierarchical structure of conceptual knowledge in the hippocampal-prefrontal system of 32 human participants (22 females). We assessed the hierarchical nature of learning during a novel tree-like categorization task via computational model comparisons. The winning model allowed to extract and quantify estimates for accumulation and updating of hierarchical compared with single-feature-based concepts from behavior. We find that mPFC tracks accumulation of hierarchical conceptual knowledge over time, and mPFC and hippocampus both support trial-to-trial updating. As a function of those learning parameters, mPFC and hippocampus further show connectivity changes to rostro-lateral PFC, which ultimately represented the hierarchical structure of the concept in the final stages of learning. Our results suggest that mPFC and hippocampus support the integration of accumulated evidence and instantaneous updates into hierarchical concept representations in rostro-lateral PFC.SIGNIFICANCE STATEMENT A hallmark of human cognition is the flexible use of conceptual knowledge at different levels of abstraction, ranging from a coarse category level to a fine-grained subcategory level. While previous work probed the representational geometry of long-term category knowledge, it is unclear how this hierarchical structure inherent to conceptual knowledge is acquired and represented. By combining a novel hierarchical concept learning task with computational modeling of categorization behavior and concurrent fMRI, we differentiate the roles of key concept learning regions in hippocampus and PFC in learning computations and the representation of a hierarchical category structure.

The Hippocampus Maps Concept Space, Not Feature Space.

The hippocampal formation encodes maps of space and a key question in neuroscience is whether its spatial coding principles also provide a universal metric for the organization of nonspatial, conceptual information. Previous work demonstrated directional coding during navigation through a continuous stimulus feature space as well as mapping of distances in a feature space that was relevant for concept learning. Here we provide the first unambiguous evidence for a hippocampal representation of the actual concept space, by showing that the hippocampal distance signal selectively reflects the mapping of specifically conceptually relevant rather than of all feature dimensions. During fMRI scanning of 32 human participants (21 females), we presented everyday objects, which had beforehand been associated with specific values on three continuous feature dimensions. Crucially, only two dimensions were relevant to prior concept learning. We find that hippocampal responses to the objects reflect their relative distances in a space defined along conceptually relevant dimensions compared with distances in a space defined along all feature dimensions. These findings suggest that the hippocampus supports knowledge acquisition by dynamically encoding information in a space spanned along dimensions that are relevant in relation to define concepts.SIGNIFICANCE STATEMENT How are neural representations of conceptual knowledge organized, such that humans are able to infer never experienced relations or categorize new exemplars? Map-like representations as supported by the hippocampal formation to encode physical space during navigation have been suggested as a suitable format. Here we provide the first evidence for a hippocampal representation of a conceptual space compared with a general feature-based space.

Improving audio-visual temporal perception through training enhances beta-band activity.

Multisensory integration strongly depends on the temporal proximity between two inputs. In the audio-visual domain, stimulus pairs with delays up to a few hundred milliseconds can be perceived as simultaneous and integrated into a unified percept. Previous research has shown that the size of this temporal window of integration can be narrowed by feedback-guided training on an audio-visual simultaneity judgment task. Yet, it has remained uncertain how the neural network that processes audio-visual asynchronies is affected by the training. In the present study, participants were trained on a 2-interval forced choice audio-visual simultaneity judgment task. We recorded their neural activity with magnetoencephalography in response to three different stimulus onset asynchronies (0 ms, each participant's individual binding window, 300 ms) before, and one day following training. The Individual Window stimulus onset asynchrony condition was derived by assessing each participant's point of subjective simultaneity. Training improved performance in both asynchronous stimulus onset conditions (300 ms, Individual Window). Furthermore, beta-band amplitude (12-30 Hz) increased from pre-compared to post-training sessions. This increase moved across central, parietal, and temporal sensors during the time window of 80-410 ms post-stimulus onset. Considering the putative role of beta oscillations in carrying feedback from higher to lower cortical areas, these findings suggest that enhanced top-down modulation of sensory processing is responsible for the improved temporal acuity after training. As beta oscillations can be assumed to also preferentially support neural communication over longer conduction delays, the widespread topography of our effect could indicate that training modulates not only processing within primary sensory cortex, but rather the communication within a large-scale network.

The Hippocampus Encodes Distances in Multidimensional Feature Space.

The hippocampal formation encodes maps of the physical environment [1-5]. A key question in neuroscience is whether its spatial coding principles also provide a universal metric for the organization of non-spatial information. Initial evidence comes from studies revealing directional modulation of fMRI responses in humans [6, 7] during navigation through abstract spaces and the involvement of place and grid cells in encoding of non-spatial feature dimensions [8]. However, a critical feature of a map-like representation is information about distances between locations, which has yet only been demonstrated for physical space [4, 9]. Here, we probe whether the hippocampus similarly encodes distances between points in an abstract space spanned by continuous stimulus-feature dimensions that were relevant to the acquisition of a novel concept. We find that, after learning, two-dimensional distances between individual positions in the abstract space were represented in the hippocampal multivoxel pattern as well as in the univariate hippocampal signal as indexed by fMRI adaptation. These results support the notion that the hippocampus computes domain-general, multidimensional cognitive maps along continuous dimensions.

Synaesthesia or vivid imagery? A single case fMRI study of visually induced olfactory perception.

The most common form of synaesthesia is grapheme-colour synaesthesia. However, rarer forms of synaesthesia also exist, such as word-gustatory and olfactory-gustatory synaesthesia, whereby a word or smell will induce a specific. In this study we describe a single individual (LJ) who experiences a concurrent olfactory stimulus when presented with congruent visual images. For some visual stimuli, he perceives a strong and automatic olfactory percept, which has existed throughout his life. In this study, we explore whether his experiences are a new form of synaesthesia or simply vivid imagery. Unlike other forms of synaesthesia, the concurrent odour is congruent to the visual inducer. For example, a photograph of dress shoes will elicit the smell of leather. We presented LJ and several control participants with 75 images of everyday objects. Their task was to indicate the strength of any perceived odours induced by the visual images. LJ rated several of the images as inducing a concurrent odour, while controls did not have any such percept. Images that LJ reported as inducing the strongest odours were used, along with colour-matched control images, in the context of an fMRI experiment. Participants were given a one-back task to maintain attention. A block-design odour localizer was presented to localize the piriform cortex (primary olfactory cortex). We found an increased BOLD response in the piriform cortex for the odour-inducing images compared to the control images in LJ. There was no difference in BOLD response between these two stimulus types in the control participants. A subsequent olfactory imagery task did not elicit enhanced activity in the piriform cortex in LJ, suggesting his perceptual experiences may not be based on olfactory imagery.