No useful theory of biological neural computation yet
I was very surprised to learn that neuroscientists have no useful theory of biological neural computation yet. I recommend reference [4] for a review. In addition to that paper, here’s few quotes from other authorities:
“How do large populations of neurons give rise to behavior? But we currently have no useful theory of biological neural computation to organize our thinking about this question or to guide the design of new experiments.” [1]
“It is useful to illustrate some of the conceptual difficulties in determining how the brain operates with an analogy to computers and electronic circuits. In your personal computer, software determines what computations are done; the underlying micro-circuitry of the computer hardware does not change when you switch from browsing the web to word processing. Is that how the brain works? Most evidence suggests that it does not. Rather, the circuits of neurons that are used, say, for recognizing speech, have been specialized through evolution to have circuitry tailored to the specific forms of computation inherent in speech recognition. These circuits are different from those used for coordinating muscles when playing tennis. In other words, circuitry in the brain seems to be special purpose hardware (rather than software) tailored to specific tasks. Given this specialization of brain “hardware” we need to understand the relationship between specific forms of micro-circuitry and specific forms of brain function. However, even though the micro-circuitry is specialized, there are likely to be standard circuits that are used over and over in different ways, the same way that certain standard circuits used in electronics for many different purposes. Thus, the emerging hypothesis is that there are certain standard microcircuits that are organized in different ways, through different synaptic wiring diagrams, to build special-purpose neural circuits. At this time, we have no idea what these standard circuits are. It is, however, clear that we must create quantitative wiring diagrams, and understand the properties of the neurons. We can then match the wiring diagram with the particular tasks performed by that portion of the brain. ” [2]
“Profound insights into neural mechanism have sprung from theory-based research. Some of the more powerful examples are the Hodgkin-Huxley model of action potential propagation, Hebbian-based plasticity rules, Barlow’s efficient coding hypothesis and Marr’s three levels of analysis. Despite these successes, there remains a disconnect between theory and experiment in biological science that would be inconceivable in the physical sciences. Technological advances render this divide even more prescient in modern-era neuroscience. It is precisely the ability to generate so much data that requires incisive experimental design and focused questions to gain understanding. As conceptualized in the cover of this issue, without this grounding, experimental neuroscience amounts to not much more than one observation heaped on another, with nearly the same utility and pointlessness that turtles bring to cosmology.” [3]
[1] https://www.simonsfoundation.org/2018/08/03/systems-neuroscience-is-about-to-get-bonkers/
[3] https://www.nature.com/articles/nn.4261
[4] https://onlinelibrary.wiley.com/doi/pdf/10.1111/cogs.12012
Author’s articles on physics and philosophy: sureshemre.wordpress.com