Nowadays, philosophers and scientists tend to agree that, even though human and artificial intelligence work quite differently, they can still illuminate aspects of each other, and knowledge in one domain can inspire progress in the other. For instance, the notion of “artificial” or “synthetic” phenomenology has been gaining some traction in recent AI research. In this paper, we ask the question: what (if anything) is the use of thinking about phenomenology in the context of AI, and in particular machine learning? (...) We will isolate one sense of “phenomenology”, namely the sense in which it is commonly understood within analytic philosophy of perception. Then, we will give examples of projects within sensory substitution and restoration science that rely heavily on machine learning and to which, according to us, phenomenology in the sense specified makes a relevant contribution. Finally, we will shed some light on what this contribution looks like and why it is important. (shrink)

According to ‘orthodox’ representationalism, perceptual states possess constitutive veridicality (truth, accuracy, or satisfaction) conditions. Typically, philosophers who deny orthodox representationalism endorse some variety of anti-representationalism. But we argue that these haven’t always been, and needn’t continue to be, the only options. Philosophers including Descartes, Malebranche and Helmholtz appear to have rejected orthodox representationalism while nonetheless endorsing perceptual representations of a fundamentally practical kind not captured by orthodox representationalism. Moreover, we argue that the perceptual science called on by contemporary philosophers to (...) defend orthodox representationalism instead motivates a return to this older view, and we suggest that contemporary philosophers may conceptualize fundamentally practical perceptual representations as _‘de agendo’_ representations, a species of representation that has constitutive appropriateness rather than veridicality conditions. (shrink)

Prima facie, accounts of scientific representation should illuminate how models support justified surrogative reasoning while remaining neutral on the nature of inductive inference. We argue that doing both at once is harder than it first appears. Accounts like “DEKI,” which distinguish justified and unjustified surrogative inferences by appealing to a distinction between derivational and factual correctness, cannot accommodate non-formal, non-rule-based accounts of inference such as John Norton’s material theory of induction. In contrast, a recent expressivist-inferentialist account appears compatible with material (...) inference, but at the cost of abandoning inductive neutrality. (shrink)