The physicist and philosopher Hermann von Helmholtz, whose foundational contributions to physics you all know very well, also invented “psychophysics,” a paradigm for phenomenology by • inferring minimal predictive quantitative models. • combining physical calculations and behavioral studies to establish that human color vision is trichromatic. • detailed mechanisms were confirmed a century later by electro/neurophysiological, genetic/genomic and biochemical methods. Now phenomenology can drive development of novel AI/machine learning technology by elucidating the neuroscience, principles and computational algorithms of cognitive processing in humans and animals. Think of, for example, generalized adversarial networks [GAN], the state of the art in deconstructing human visual processing.
Cephalopods (octopus, cuttlefish, and squid) modulate their skin color and texture to match their marine backgrounds on millisecond timescales; in this sense, they “report” to us their perceptions. I will describe how biologists in my unit at OIST have developed a flexible model organism ideal for deconstructing this process. Now my unit is pursuing the development of customized physical hardware and software to quantitatively probe the i/o response function of this model organism, and to reverse engineer the biophysics and neural algorithms of its cognition and consciousness.