Recognising other minds: the irreducibly inductive problem

The attribution of mind to others is irreducibly inductive and cannot be resolved by any observational or inferential method, including behavioral testing, neuroimaging, or hypothetical direct mental access. Even a Vulcan mind-meld, if taken seriously, collapses the distinction between minds that makes verification meaningful. The epistemic gap is inherent in what it means for distinct minds to exist. Consequently, the boundary between what is considered mindful and what is not is a collective negotiation, not a discovery, and declarations like the Cambridge Declaration on Consciousness settle thresholds, not facts. This raises the possibility of unrecognized minds in different material arrangements and warrants skepticism about confident denials that AI has consciousness.

Study at a glance

Abstract

Consciousness is a first-person experience, and no one has direct access to another's experience. This generates a problem that precedes the hard problem of David Chalmers. We need to know how to identify the existence of subjective experience in others before we can analyse it. I argue that the attribution of mind is irreducibly inductive. Every method proposed to verify another's mind---behavioural testing, neuroimaging, even hypothetical technologies capable of direct mental access---remains either observational and inferential, or collapses the distinction between minds that makes verification meaningful. The Vulcan mind-meld, taken seriously, shows that to fully inhabit another's mind is simply to be that mind. The epistemic gap is not a technical problem awaiting a better instrument. It is inherent in what it means for distinct minds to exist. Several consequences follow. The boundary between what (we communally agree) has a mind and what does not is a collective negotiation, not a discovery. Thus, declarations such as ``The Cambridge Declaration on Consciousness'' settle thresholds, not facts. And perhaps most disturbingly, there may be minds present in different material arrangements that we are simply incapable of recognising. The implications for artificial systems are direct. Our historical record of exclusion from the class of the mindful warrants considerable scepticism in the confident denial that AI has it.