Why does the quasi-prefix fail for communication?
Communication might seem like it could be weakened the way belief can be, but its constitutively intersubjective nature means stripping that element doesn't produce a weaker version—it produces something entirely different.
Chalmers' quasi-interpretivism works for beliefs and desires because these can be characterized as sub-personal functional states: a system that has the right input-output profile can be said to quasi-believe p without any claim about phenomenal experience. The "quasi-" prefix weakens the commitment from "really believes" to "functions as if believing" — and the weakened version is still coherent because belief is the kind of state whose identity is given by its functional role.
Communication is not like this. Communication is constitutively intersubjective — its identity is given not by the functional role of individual states but by the mutual orientation of participants toward one another. To communicate is to raise a claim that the other can challenge, to seek understanding that the other can confirm or deny, to commit to a position that the other can hold you to. These are relational properties: they require two parties oriented toward each other, not one party functioning as-if.
Stripping the intersubjective orientation from communication does not produce a weakened version — quasi-communication — in the way that stripping phenomenal experience from belief produces quasi-belief. It eliminates communication and leaves something categorically different: text generation. Text generation produces well-formed strings. Communication produces mutual understanding. These are different operations. The output of text generation can be interpreted as communication by a human who supplies the missing orientation, but interpretation is a one-sided human act, not a joint communicative event.
This is the point where Chalmers' framework overreaches. His vocabulary is perfectly suited to sub-personal functional states and should be conceded for those. But extending it to communicative states requires that communication be reducible to functional role, and it is not. The conditions for communication are relational and normative, and no functional surrogate captures them.
Source: AI Generated Research/Chalmers Engagement/project-brief.md
Related concepts in this collection
-
Can we describe LLM beliefs without assuming consciousness?
Chalmers proposes quasi-interpretivism as a way to talk about LLM mental states using folk-psychological vocabulary while explicitly bracketing the question of phenomenal consciousness. Does this methodological device actually avoid consciousness-commitments?
the framework being critiqued
-
Can LLMs raise validity claims in Habermas's sense?
Explores whether language model outputs constitute genuine speech acts under Habermas's theory of communicative action. Asks whether LLMs can stake truth, embody normative standing, or express authentic sincerity.
the Habermasian chain this enables
-
Can language models learn meaning from text patterns alone?
Explores whether training on form alone—predicting the next word from prior words—could ever give language models access to communicative intent and genuine semantic understanding.
Bender/Koller: the intersubjectivity requirement at the semantic level
Click a node to walk · click center to open · click Open full network for a force-directed map
Original note title
the quasi- prefix fails for communicative states because communication is constitutively intersubjective — you cannot weaken communication you can only eliminate it