This dissertation brings together a collection of four projects that are thematically related through their relevance to the ”symbol-grounding problem” in cognitive science: the issue of how the internal activity of a cognitive agent (i.e. neural activity at a biological level, or ”representations” at a cognitive level) are meaningfully connected to things in the world. This is a general problem for theories of cognition, which must be solved to have adequate theories of language more specificall—linguistic meaning is not possible unless we can explain how meaning is possible in general. I begin in chapter one with an overview of the symbol-grounding problem, situated in the debate between computational (representational) theories of cognition and non-representational theories such as ecological psychology, enactivism, and embodied cognition. In chapter two, I present a computational model of ”neural resonance,” which offers an account of the representational role of neural activity that does not require thinking of representations as ”encodings” of things in the world, and therefore may not fallprey to the symbol-grounding problem. I show how simple homeostatic mechanisms at the level of neurons may give rise to transient localist representations that can control the action-perception loop of an agent, and also leads to emergent prediction-like behaviors in language processing. In chapter three, I present two human subjects experiments that investigate the possibility that language comprehension is grounded in sensorimotor simulation. Abstract or metaphorical language is a critical test case for this hypothesis, and I report evidence that such language does not generally depend upon sensorimotor simulation, but that literal language does not always depend upon it; rather, we observe a continuum of sensorimotor involvement in language processing. In chapter four, I report the results of another human-subjects study on the context-flexibility of phonetic representations in Spanish-English bilinguals. This experiment provides support for the notion that perceptual categories crucial for language processing can be flexibly adapted to fit the current context. In chapter five, I present an agent-based model of ”collective cognitive alignment” which addresses a crucial step in the emergence of language: the coordination of shared perceptual categories. Finally, in the concluding chapter I reflect on how we may construct a theory of cognitive science that takes meaning seriously, and allows us to preserve an unbroken ”grounding wire” as our theories move from the cognitive activity of the simplest lifeforms up to the most complex forms of cognition as exemplified in human language and culture.