|
LLMs and the Problem of Uncommon Ground
Following the Symbol Grounding Problem, it is generally thought that any natural language system, in order to realize meaningful outputs, must be dependent on some other system or ground. Given their near-flawless natural language outputs, some have suggested that Large Language Models (LLMs) may possess some form of semantic “understanding” and are thus likely to be grounded in some way. I support this view and maintain that LLMs are referentially grounded. However, since this kind of grounding significantly differs from the sensorimotor grounding of human language systems, I argue that communication between both systems will lead to divergences. Crucially, these divergences will fundamentally limit reliability and mutual understanding between humans and LLMs. I support this claim by introducing research on neurosemiotics and embodied semantics, which demonstrates that much of our semantic content is fundamentally grounded in worldly interactions and sensory responses. Ultimately, I contend that critical differences in grounding form a persistent barrier that will deny two differently grounded language systems from achieving reliable compatibility. Neuroaesthetics: A Primer for Philosophers
Abstract: Neuroaesthetics initially developed as a branch of cognitive neuroscience that studied neural responses to artworks and aesthetic experiences. Recent research, however, reveals a maturing science that is expanding into a science of sensory preferences and hedonics. On this account, neuroaesthetics studies the neural correlates responsible for the realization of sensory responses, as well as how and why these responses attain hedonic value. While neuroaesthetics has garnered interest from other branches of neuroscience, its significant developments are largely unknown to other disciplines. The goal of this project is to bridge this epistemic gap by introducing neuroaesthetics to a wider philosophical audience. I provide an account of its brief yet noteworthy history and, importantly, demonstrate that neuroaesthetics, since it is now a science of hedonics and sensory valuations, can be of interest to philosophers working on a range of issues. Reframing the Hard Problem of Consciousness as the Hard Problem of Aesthetic Experience
Abstract: The hard problem of consciousness represents the challenge of elucidating how and why physical brain processes give rise to rich qualitative and subjective experiences. In this paper, I introduce accounts of aesthetic experiences developed by neuroaesthetics researchers and highlight that the sensory and evaluative properties typically attributed to phenomenal consciousness mirror these empirical accounts of aesthetic experiences. Thus, in this view, the hard problem of consciousness can also be thought of as the hard problem of aesthetic experience. However, even though these concepts mirror one another, I argue for the replacement of phenomenal consciousness with that of aesthetic experience. I defend this position due to empirical accounts of aesthetic experiences being less ambiguous, their neural correlates and functions have largely been established, and thus their proximate and ultimate explanations do not require metaphysical speculations that extend beyond the reach of the material sciences. How Solving a Dopamine Puzzle Might Resolve the Paradox of Painful Art
Abstract: Aesthetic hedonism, the view that the value of an artwork is determined by its pleasure-inducing properties, is largely regarded as the default theory of aesthetic value. However, aesthetic hedonism has significant challenges as well. One notable challenge points to the high value we ascribe to some works of art that arouse negative or unpleasant sensations, like Picasso’s Guernica. Relatedly, the paradox of painful art points out that some of us seek out art that evokes unpleasant sensations yet avoid unpleasant sensations in real life. If this claim is correct, then it appears that aesthetic hedonism is false. With this project, I draw a parallel between the paradox of painful art and a neurobiological puzzle (the cocaine puzzle), which asks how cocaine induces hedonic sensations even though it does not directly activate hedonic neural mechanisms. I ultimately demonstrate that the neural mechanisms that can resolve the cocaine puzzle can also resolve the paradox of painful art. Importantly, this response to the paradox of painful art also defends aesthetic hedonism since it highlights underlying hedonic mechanisms that become active even in the face of painful art. Expanding a New Mechanical Ontology: Objects as Functional Mechanisms
Abstract: According to several New Mechanists, most phenomena (natural or otherwise) are actively brought about by mechanisms. However, some have suggested that the phenomenon a mechanism is responsible for could be a disposition or a capacity to realize a function. If this latter account is accepted, then it appears that some inactive objects can be considered mechanisms. This inclusion would be noteworthy since inactive objects have typically been denied mechanism status within a new mechanical ontology. With this paper, I will defend the position that inactive objects, which have a capacity to realize a function, are mechanisms (hereafter individual mechanisms). I will argue that this type of mechanism should coexist with the received view that mechanisms underlie or produce some phenomenon (hereafter process mechanisms). In my view, accounting for objects within a new mechanical ontology will expand and clarify the deep role mechanisms play within the causal and physical structure of the world. |