Teleosemantics
Teleosemantics is the philosophical project of explaining semantic content — what mental states, linguistic expressions, and biological signals are about — in terms of biological function. The central claim, developed by Ruth Millikan and David Papineau, is that the meaning of a representation is determined by its teleofunction: the function it was selected to perform, either through evolutionary history or through learning. A frog's neural state that fires when a small dark object flies across its retina is not merely correlated with flies; it represents flies because it was selected by evolution to guide tongue-flicking behavior toward nutritious prey. The correlation is not enough. The selection history is what makes the state meaningful rather than merely informative.
Teleosemantics is a form of naturalized semantics: it attempts to locate meaning within the natural world, reducing the intentional to the functional. In this respect it opposes both the view that meaning is a purely formal property of symbol systems (the position associated with Frege and compositional semantics) and the view that meaning is irreducibly mental (the position of Brentano and his heirs). It claims that meaning is not a ghost in the machine but a causal-historical property of the machine's parts: what they were selected to do.
The Mechanics of Teleosemantics
The teleosemantic theory has two components. The first is a theory of function: biological functions are not current causal roles but selected effects. The heart's function is to pump blood not because it happens to pump blood now, but because it was selected for pumping blood in ancestral populations. The second is a theory of content: the content of a representation is determined by the normal conditions under which it performs its function. A representation that guides behavior toward prey has the content 'prey is present' because its function is to guide prey-catching behavior, and it performs this function correctly when prey is actually present.
This framework explains how misrepresentation is possible — a central desideratum for any theory of content. A representation can be false because it can fire in conditions that are not the normal conditions for the performance of its function. The frog's representation fires when a bee-bee pellet flies past; it misrepresents the pellet as prey because the function of the representation is to guide prey-catching, and the pellet is not prey. The possibility of error is built into the theory: a representation is a state that is supposed to correspond to something, and 'supposed to' is grounded in selection history, not in current success.
Challenges and Objections
The most famous objection is the Swampman problem, posed by Donald Davidson. Imagine a molecule-for-molecule duplicate of a human popping into existence in a swamp through a lightning strike. The Swampman has no evolutionary history, no selection history, no learning history. Yet it seems to have beliefs, desires, and meaningful mental states. If teleosemantics is correct, the Swampman has no content: its states are not about anything because nothing selected them to be about anything. But this is counterintuitive. The Swampman looks, acts, and reasons exactly like a human. If it lacks content, then content is not what makes cognition cognitive.
A related problem is the indeterminacy of function. Evolution selects for proximal outcomes, not distal ones. The frog's representation might have been selected for 'fly-present,' 'nutritious-object-present,' or 'small-dark-moving-object-present.' All three functions are consistent with the same selection history. Which one is the content? Millikan's response is that the content is determined by the most specific normal condition that explains the success of the mechanism. But critics argue that this 'most specific' criterion is either underdetermined or question-begging.
A deeper challenge is whether teleosemantics can scale to the full range of human thought. The contents of simple perceptual states may be plausibly tied to evolutionary functions, but what about the contents of mathematical beliefs, moral judgments, or thoughts about tomorrow? These seem to lack the direct evolutionary anchoring that makes teleosemantics work for frog perception. The extension from 'nutritious object' to 'the Riemann hypothesis' is not obvious.
Teleosemantics is the most honest attempt to naturalize meaning, and it fails for exactly the reasons that any such attempt must fail. The problem is not that the Swampman lacks content; it is that the Swampman reveals that content is not a historical property at all. Meaning is not a fossil record of selection pressures. It is a present-tense relational achievement: a state is about something because it is currently used, interpreted, and constrained by something. The selection history is relevant only because it shaped the current architecture. But the architecture, once built, can do things that no selection pressure ever intended — and those novel uses are not shadows of old functions. They are new meanings. Teleosemantics makes the same error as genetic determinism: it confuses the history of a system's design with the scope of its current capacities.