Fluid semantics: Semantic knowledge is experience-based and dynamic

Abstract

Is our internal notion of, e. g., the object lemon, static? That is, do we have stable semantic representations that remain constant across time? Most semantic memory researchers still (at least tacitly) take a static perspective, assuming that only effects that can be demonstrated across a variety of tasks and contexts should be considered informative about the architecture of the semantic system. This chapter challenges this perspective by highlighting studies showing that the cognitive and neural representations of object concepts are fluid, changing as a consequence of the context that each individual brings with them (e. g., via current goals, recent experience, long-term experience, or neural degeneration). These findings support models of semantic memory in which rather than being static, conceptual representations are dynamic and shaped by experience, whether that experience extends over the lifetime, the task, or the moment.

Semantic memory: A review of methods, models, and current challenges

“…Furthermore, Yee, Huffstetler, and Thompson-Schill (2011) used a visual eye-tracking paradigm to show that as an object unfolds over time (e.g., auditorily hearing frisbee), particular features (e.g., form-related) come online in a temporally constrained fashion and can influence eye fixation times for related words (e.g., e.g., participants fixated longer on pizza, because frisbee and pizza are both round). Taken together, these findings suggest that semantic memory representations are accessed in a dynamic way during tasks and different perceptual features of these representations may be accessed at different timepoints, suggesting a more flexible and fluid conceptualization (also see Yee, Lahiri, & Kotzor, 2017) of semantic memory that can change as a function of task. Therefore, it is important to evaluate whether computational models of semantic memory can indeed encode these rich, non-linguistic features as part of their representations.…”

Events as intersecting object histories: A new theory of event representation.

“…An intuition for how such representations can be individuated, given the recurrent and relational processes described above, can be conveyed as follows: For objects that we have experienced before, or which are instances of types we have experienced before, the semantic memory activated by the object’s sensorimotoric features reinforces just those features in the perceptual experience that map onto that memory—the semantic type corresponding to cats, activated through experiencing Garfield, becomes a part of the experience of Garfield, distinguishing that experience from the experience of Odie, or of the pie, or of Jon Arbuckle’s legs. These higher-level representations that are coactivated with the sensorimotoric features from the episodic experience are to those features as the puppeteer’s strings are to the marionette—they individuate the body of one marionette from the body of another (the relationship between the position of the marionettes and the configuration of those strings changes dynamically through time, in much the same way as conceptual activation, and the relationship between semantic memory and the sensorimotoric input activating that input, is itself dynamic; see Yee, 2017 for review). For objects we have not experienced before and which we encounter for the first time (i.e., for which we have no higher-level semantic or other memory that is coactivated with the perceptual experience), features that travel together through space and time are self-reinforcing—for example, Garfield’s paws (if we had never seen an animal before) travel with his arms across our experience of Garfield more than they do with pies.…”

Can the Machine Understand: An Evidence Based Approach to the Chinese Room

“…Language learning is a complex process spanning multiple developmental stages as well as many faculties (Bjorkland and Causey, 2017). While native arguments such as universal grammar have been poised and are still under much debate (Yang, 2004), it would seem that much of language is derived from experience (Yee, 2017). This can occur very early with children learning the language of which they are raised and exposed to: A child raised in China would learn Chinese.…”