Neurocognitive insights on conceptual knowledge and its breakdown

Matthew A. Lambon Ralph 0 0 Neuroscience and Aphasia Research Unit, School of Psychological Sciences, University of Manchester , Zochonis Building, Brunswick Street, Manchester M13 9PL , UK Articles on similar topics can be found in the following collections Receive free email alerts when new articles cite this article - sign up in the box at the top right-hand corner of the article or click here References Subject collections Email alerting service rstb.royalsocietypublishing.org Review Cite this article: Lambon Ralph MA. 2014 Neurocognitive insights on conceptual knowledge and its breakdown. Phil. Trans. R. Soc. B 369: 20120392. http://dx.doi.org/10.1098/rstb.2012.0392 Subject Areas: cognition, neuroscience Author for correspondence: Matthew A. Lambon Ralph e-mail: Neurocognitive insights on conceptual knowledge and its breakdown Matthew A. Lambon Ralph Conceptual knowledge reflects our multi-modal semantic database. As such, it brings meaning to all verbal and non-verbal stimuli, is the foundation for verbal and non-verbal expression and provides the basis for computing appropriate semantic generalizations. Multiple disciplines (e.g. philosophy, cognitive science, cognitive neuroscience and behavioural neurology) have striven to answer the questions of how concepts are formed, how they are represented in the brain and how they break down differentially in various neurological patient groups. A long-standing and prominent hypothesis is that concepts are distilled from our multi-modal verbal and non-verbal experience such that sensation in one modality (e.g. the smell of an apple) not only activates the intramodality long-term knowledge, but also reactivates the relevant intermodality information about that item (i.e. all the things you know about and can do with an apple). This multi-modal view of conceptualization fits with contemporary functional neuroimaging studies that observe systematic variation of activation across different modality-specific association regions dependent on the conceptual category or type of information. A second vein of interdisciplinary work argues, however, that even a smorgasbord of multi-modal features is insufficient to build coherent, generalizable concepts. Instead, an additional process or intermediate representation is required. Recent multidisciplinary work, which combines neuropsychology, neuroscience and computational models, offers evidence that conceptualization follows from a combination of modality-specific sources of information plus a transmodal hub representational system that is supported primarily by regions within the anterior temporal lobe, bilaterally. 1. Introduction Semantic cognition refers to a collection of interactive cognitive mechanisms that support semantically derived behaviours. We use our semantic or conceptual knowledge not only for verbal comprehension but also when we initiate language production (the purpose of receptive and expressive communication is, after all, the transfer of meaning from the speaker/sender to the listener/receiver). In addition, our considerable database of semantic knowledge is crucial in the non-verbal domain, both receptively (identification of non-verbal stimuli necessitates the transformation of sensation to meaning) and expressively (drawing and other expressive arts are based on the transmission of meaning, whilst effective object use requires semantic knowledge of each implement). Semantic cognition can be decomposed into three interactive principal components underpinned by separable neural networks: (i) semantic entry/exit, i.e. translation between sensation/motor representations and semantic knowledge; (ii) the long-term representation of concepts/semantic memory; and (iii) semantic controlmechanisms that interact with our vast quantity of semantic knowledge in order to generate time- and context-appropriate behaviour [1,2]. Every semantic task (receptive or expressive) requires a variable combination of all three components. Consequently, when any one of them is compromised & 2013 The Authors. Published by the Royal Society under the terms of the Creative Commons Attribution License http://creativecommons.org/licenses/by/3.0/, which permits unrestricted use, provided the original author and source are credited. (after neurological damage or transient brain stimulation), participants will fail in semantic assessments though the quality of their impairment will vary. This review is focused primarily upon semantic representationthat is the nature of coherent concepts, how they are represented and their neural basis. A brief detour into the nature of semantic entry/exit and control provides important information not only with regard to what each of these principal components of semantic cognition is, but also what semantic representation is not. In advance, however, it is worth underlining the observation that these three principal components have to be highly interactive in order to support semantic activities. Specifically, variation in efficiency within each system (either because the stimuli/concepts/contexts are inherently challenging or because a component has become compromised) will lead to automatic up- or downregulation of contributions from the other components. For example, the uncertainty that follows from noisy stimuli can be compensated by upregulating the bidirectional interaction with meaning (i.e. the semantic representations) or context. Likewise, there will be variable involvement of the three components depending on the nature and demands of the task or concept (e.g. for a formal exploration of this issue with respect to concrete and abstract concepts, see Hoffman et al. [3]). Dedicated cognitive and neural machinery is taken up with semantic entry, i.e. reception of sensation and its translation into meaning, and also with semantic exit that is the transformation of meaning into the motor sequences that allow us to express our knowledge to others (e.g. through speech, writing, drawing, etc.). Each sensory motor domain requires modalityspecific computations that are necessary for transformation of sensation and these are supported by different cortical and subcortical regions and pathways. These sensory-specific processes are observed not only in functional neuroimaging studies but also through the modality-specific disorders exhibited by some neurological patients. Lissauer ([4] Jackson translation) was one of the first researchers to note that within the visual domain, there is a clear separation of patients with damage to the primary visual machinery (generating apperceptive agnosia) versus other patients with deficits in higher-order semantic representations (associative agnosia). Parallel distinctions are found in the other sensory domains with regard both to intact function (as revealed by in vivo neuroimaging) and neuropsychological studies [5]. The crucial distinction between entry/exit processes and core semantic representati (...truncated)