Niranjan Krishna

These are working essays for the ongoing Linguistic Epistemology project. The linguistic terminology used will be Saussure’s.

---

The Hallucinatory Sign – When grounding linguistic certainty, an obstacle that hampers progress is the Hallucinatory Sign. Consider the conversation of two Agents – A and B. Imagine B becomes receptive to a signifier S_i from A. However, upon further enquiry, A denies producing the signifer.

The case of pseudo-hallucinations is irrelevant here; for instance, that A lied to B. But, what about a real hallucinatory case? A doesn’t produce a signifer S_i, however, B receives one anyway. If anything, this shows the capability of an unperceived sign producer and doesn’t hold contention to the epistemic certainty of language itself. Language, and its certainty, is held with the fact that S_i does represent some Signified S_d. 

---

Sign-Producing Agents – Humans as sign-producing agents, a hint on why the Turing test is designed the way it is. However, in no way are Large Language Models sign-producing agents, as they still cannot produce out of no prompts. But that’s a tangent.

The case of the hallucinatory sign can be shed light upon through psychoanalysis. Unconscious as a sign-producing agent. Our senses aren’t concerned with explicit representation of “objective” reality, even the word remains incoherent to speak of. Information is mediated before perception. Then it’s not far-fetched to say that the unconscious can mediate signs, alter them and even generate ones.

A point of interest is animals as sign-producers. We do see primitive sign-production capabilities from animals. However, one does discern a difference in complexity between the sign-production, and the subsequent meaning-making, that humans exhibit in comparison. More to be studied.

---

Law of Identity in Language – Logic is predicated on its laws, the first one being A=A: the Law of Identity. In an amalgamated Kantian-Wittgensteinian spirit, the tautological a priori that is the law of identity doesn’t interest me. But, what is the maximal a posteriori law of identity? That, I claim, is found in language. 

In Truth, Certainty and Language, we conclude that the maximal epistemic certainty possible belongs to that of language.  That in a Language L, between two agents (for simplicity), the signifier S_i represents the same signified for both. Formally, say the common signifier is S_i and the signified for A is S_dA, for B S_dB. Then, the maximal a posteriori law of identity becomes

S_dA = S_dB

Note, that the law of identity persists, and can only persist, in some Language L; and thereby bound to its context. Thus, eliminating the possibility of any “objective” truth, as all truth, and epistemic certainty alike, becomes contextual. Our search becomes coherence within that context – perhaps a new logic has to be invented?

References

Kant, Immanuel. Critique of Pure Reason. 1781.

Krishna, Niranjan. Truth, Certainty and Language. 2023.

Saussure, Ferdinand de. Course in General Linguistics. 1916.

Wittgenstein, Ludwig. Philosophical Investigations. 1953.