Is a Diamond More Elegant than a Diamond?: The Role of Sensory-Grounding in Conceptual Content Chelsea L. Gordon (cgordon7@ucmerced.edu) University of California at Merced, Department of Cognitive Science, 5200 North Lake Road, Merced, CA, 95343 Sarah E. Anderson (sec57@cornell.edu) Nielsen Company, 53 Brown Rd. Ithaca, NY 14850 Michael J. Spivey (spivey@ucmerced.edu) University of California at Merced, Department of Cognitive Science, 5200 North Lake Road, Merced, CA, 95343 Abstract It has recently been suggested that much of the research in embodied cognition can be explained by a “disembodied” account in which conceptual and cognitive processes perform their computations in a modular fashion and the sensory and motor associations that show up in embodiment experiments may arise merely from spreading activation from the cognitive module to the sensory and motor systems (Mahon & Caramazza, 2008). In such a model, the cognitive module processes its information and accesses its representations exactly the same way as it always would have, and the embodiment effects are essentially epiphenomenal. We test this idea by manipulating the sensory aspects of the perceptual input that triggers the activation of a concept. Throughout the history of conceptual representation research, feature lists of concepts have been treated as a method for accessing the semantic content of those conceptual representations. When there are sensory differences in the font of the written word that triggers accessing of a concept, does the concept get accessed in a different way? Are different conceptual features more prominent than others? We find a series of conceptual features that are more prominent when the concept is presented in one font versus the other. Continuations of this research project involve reaction-time priming experiments to see if these differential access effects happen at the timescale of hundreds of milliseconds. Our results are discussed in the context of competing or compatible accounts of embodied and symbolic cognitive processes. Keywords: language, embodied cognition, conceptual representation, psycholinguistics Introduction Converging evidence in the field of embodied cognition suggests that low-level perceptual processes, such as auditory and visual processing, are activated immediately and automatically during higher-level tasks, such as language and conceptual processing (Barsalou 1999; Bergen, Lindsay, Matlock, & Narayanan 2007; Calvo- Merino, Grezes, Glaser, Passingham, & Haggard 2006; Thelen & Smith 1994; Zwaan 2004). There remains debate, however, about the directness of this interaction. For instance, traditional cognitive science predominantly responds to findings in embodied cognition by suggesting that the core of the phenomenon is a symbolic representation and that after this process becomes active, the activation merely spreads to sensory and motor associations (Mahon & Caramazza, 2008). This view then implies that sensorimotor groundings are peripheral, not central, to conceptual representations. Furthermore, they argue that embodied cognition could only be plausible in regard to concrete concepts, since for abstract concepts such as “justice”, it can be difficult to identify sensory and motor information that grounds it to the world. However, concepts are not spontaneously activated from inside. Activating concepts, like flower, typically requires the environment to deliver a sensory stimulus to the observer. This stimulus will unavoidably have idiosyncratic sensory properties that must be completely discarded in order for the exact same symbolic concept (of flower) to get activated in exactly the same way every time one sees a flower or hears the word “flower”. The case is the same for concepts such as “justice”. Whether reading the word, hearing the word, or thinking the word, there is a contextual situation made up of sensory information that brought the concept to mind. Moreover, evidence suggests that participants do not summarily discard such idiosyncratic information, as is seen in the phonemic categorical perception literature. For over a decade, the evidence indicated that the process of phoneme discrimination was discretely categorical (Liberman, Harris, Kinney, & Lane, 1961). Take for example the categorical perception of the phonemes “pah” and “bah”. These two phonemes differ only on the dimension of voice onset time (VOT): if VOT is between 0 and 30 ms, then the sound is perceived as “pah”, and if VOT is between 30 and 60 ms, the sound is perceived as “bah”. When participants are asked to decide which of the two phonemes a sound is, this difference appears discretely divided at the 30 ms boundary. More recently, further investigation of categorical perception has yielded more graded results. Just analyzing data from the endpoint of a response (such as accuracy) runs the risk of overlooking fine-grained details available before a response is fully executed (Abrams & Balota, 1991). 2293