Works in Progress

Abstract

Our inability to describe a computational solution to the hard problem of phenomenal consciousness does not preclude the existence of a solution, but it does curtail the aspirations one might entertain for engineering minds. How much else is at stake, though? In this paper I explore the extent to which the apparent hard-limit of phenomenal consciousness likewise curtails computational accounts of agency. Framed in terms more familiar to the philosophy of mind, I conclude that while philosophical zombies can be agents, they cannot be autonomous agents. The hard problem of phenomenal consciousness is thus a vicious computational contagion: Autonomous agency, moral responsibility, and personhood all join phenomenal consciousness in the quarantine of the computationally—and, perhaps, physically—intractable.

Abstract

Confronted by the hard problem of phenomenal consciousness, mysterianists council more than epistemic humility. We are, they conclude, cognitively closed to its solution, much in the way a cat is cognitively closed to a proof of the Pythagorean theorem. There is no proof the cat can ever grasp, and neither can we ever know just how subjective experience is the natural property of the brain it presumably is. While mysterianists have rightly been taken to task for asserting that our capacity to grasp solutions can become entirely unhinged from our ability to clearly formulate problems, their arguments do call attention to a curious fact: we wholly lack any introspective access to the brain states causing phenomenal consciousness--or, indeed, any brain states whatsoever. From the confines of subjective experience, the brain is a thoroughly black box, closed to the felt introspection my stubbed toe is not. Diagnoses for this inability vary. In this talk I argue that computational theories of mind, whether classical or connectionist, fully explain cognitive closure as a complexity limitation on cognition.

Abstract

The prevailing philosophical view seems to assert that fully ethical artificial agents require (at least) a trio of cognitive capacities--awareness, understanding, and free will--each of which presently exceeds either our engineering grasp or our computational modeling by no small measure. Consequently we shelve puzzles over, say, robots as fully moral agents in favor of the problem of designing moral normative constraints on their ranges of behavior. Extending recent scholarship on the arguments PF Strawson offered in his influential "Freedom and Resentment", I argue that the nature of participant reactive attitudes like resentment and gratitude are such that they apply to us regardless of whether we have awareness, understanding, or free will, and will apply to artificial agents regardless of whether they have the same or similar cognitive capacities as human persons.

Abstract

Scientific inquiry today thrives on a substrate of computational methods. On the side of theory, complex biological, chemical, meteorological, and astrophysical systems are modeled computationally. On the side of empirical observation, vast arrays of sensors accumulate data, rendering observed experimental results in terms of human perception a rather quaint notion. Our hope must be that computer science itself enjoys a privileged epistemic status akin to mathematics and logic so as to protect the integrity of computationally infused scientific inquiry. By revisiting early twentieth century arguments on the ambiguity of falsification and underdetermination of theory by the evidence in light of the computational turn in scientific practice, however, I argue that adopting computational methods in scientific inquiry poses epistemic hazards even when---and perhaps especially when---everything goes precisely as planned.

Abstract

Hume's rejection of the Principle of Infinite Divisibility has been widely derided by mathematicians and philosophers alike as, at best, evidence of his mathematical naïveté and, at worst, as a reductio on humean empiricism. In this paper I argue that Hume's rejection of the Principle is neither evidence of mathematical obtuseness nor an indictment of his empiricism. To the contrary, it implies a careful and principled rejection of the platonic conception of mathematics in light of limitations on empirical observation which, I argue, can now be seen as essentially computational in nature.

Abstract

Clark and Chalmers argued that cognition itself extends to the kinds of artifacts we use in accomplishing our ends. The classic example of this is to ask, in what lies the cognitive difference between a shopping trip flawlessly conducted from memory versus a shopping trip flawlessly conducted by following a list written on a slip of paper. From a functionalist standpoint, the externalized shopping list is as much a component of cognition as the internalized list. Thus we have the thesis of externalism. But then, astute students ask, why is our use of Large Language Models against which you've been religiously railing all semester not simply another dimension of externalized cognition? If the artifact of a slip of paper inscribed with a shopping list counts as cognition given its functionally indiscernible substitution with a memorized list, then why is a prompt put to ChatGPT, say, not similarly an externalizing a cognitive process via an admittedly vastly more complicated technology than pencil and paper? In this essay I argue that there are bounds to the thesis of externalism by arguing that the technology of generative-AI is not merely different in degree of complexity and sophistication, it is also different in kind in such a way that it cannot be viewed as simply another dimensions of an individual's cognition.

Abstract

There are many reasons why a professor might prefer their children go elsewhere: providing them with more opportunities, saving them the embarrassment of over-hearing ones own students, giving them necessary independence to forge their own transition to adulthood, and so forth. All of these reasons are clearly compatible with my own University being merely a slightly less desirable, all things considered, but nevertheless acceptable--and certainly, cheaper--alternative. What if that were not the case, however? What if my own university were in no way an acceptable alternative? What would have to be the case about the university such that it warrants striking from the list of possible alternatives? And what, further, would that say about my own continued collaboration with such an institution?