Information Processing is not dead
An article in Aeon Magazine (LINK HERE) by Robert Epstein cries out the death of compuationalism (the approach in neuroscience that makes an analogy between our minds and computers). The piece hits you early with the hard stance that “Your brain does not process information, retrieve knowledge or store memories. In short: your brain is not a computer”.
This article, in my opinion, is horribly misleading. Many cognitive scientists would agree that the brain isn’t a computer. For myself, as someone who does research at the nexus of cognitive psychology and artificial intelligence, the computational analogy is seriously misleading.
As a pithy issue to begin with, he seems to rest his argument on the work of Ray Kurtzweil, the innovative inventor and futurist. While I’m in many ways a fan of Kurtzweil, I believe his predictions are poorly estimated and that many — I’m afraid — may be more ideologically than empirically based. I believe Epstein’s argument would be stronger had he based his claims on the work of a cognitive scientist. This, I’m afraid, set’s up part of the straw man he attacks.
In any case, my main problem is that the argument entangles “information processing” and “the brain is a computer” in ways that are not very valid. While he is right that information isn’t stored like a computer and retrieved, this doesn’t make the analogy worth throwing out in its own right. Most scientific theories work by drawing analogies to some model, while admitting that there is not complete correspondence between the phenomena in the real world and the theory’s associated variables. It is the utility of the theory and the closeness of the correspondence that make a theory strong (among other things).
This issue instantiates itself in many ways. One is by not leaving the reader with a hint of what is better. While the strict computationalist approach is incompatible with many things, the idea that embodiment and interactionist approaches are not compatible with information processing is simply untrue. Work in agent based modelling and Multi-Agent AI have done just this, for example.
However, the rabbit hole here gets deep, and the more we look into how similar brains and computers are, the more issues appear for the idea that Epstein’s argument holds water.
For example, while I don’t believe the brain is a computer, even by strong analogy, denying that it can process information leaves one wondering how it is we can explain recent breakthroughs in brain computer interfaces and robotic prosthetics. I’ll list a few here with hyperlinks to resources that you may find interesting:
For example, some advanced prostheses are integrating computers with our own nervous system. And recent advances are allowing for implantable devices that measure brain signals so that we can control electronics.
Some of these systems are reportedly so advanced that they are able to reverse engineer tactile sensation effectively.
These facts make it hard not to provide at least some benefit of the doubt that the brain does process information, and does so in patterned and effectively predictable ways. Furthermore, the fact that these are prostheses show that aspects of corporeal embodiment can also be taken into account by an information processing approach, if not through the computational approach as such.
But, the rabbit hole goes further still. Recent work on growing physical neurons on computer chips has allowed for neurons to integrate themselves into control based systems that can perform advanced behaviours such as flying f-22 raptor (fighter jets) in simulation:
And creating new organs that are integral to sensation (i.e. information perception) has also been addressed by recent advances, such work is literally helping the blind to see.
And as with any science, some work is controversial and raises thorny ethical issues. Such as allowing us to exercise “remote control” actual neural circuitry in situ; this lets us exercise control over animals like rats directly with electrodes.
So yes, the computer analogy may be just an analogy, but information processing itself is quite real and research and findings such as the above, as insane as they may be at times, shows how powerful the information processing approach really is, not just in analogy, but as a basis for applied science.
From a less applied perspective, the idea that we don’t have perfect representations of things in our minds that we can recall is, to someone who studies cognition and culture, rather uninteresting. We have been dealing with these issues for decades now and have empirically shown that these flaws are interestingly patterned, and can be viewed within an information processing paradigm. Some researchers, such as Thom Scott-Phillips, even believe that they can be fit within a representationalist paradigm as well.
I think these issues call for many who would deny information processing to rethink the differences between information processing as a research paradigm for understanding a wide range of phenomena from a cognitive perspective and the computational analogy. The computational analogy is used in the most basic sense in cognitive science, and no respectable cognitive scientist I’m aware of denies Epstein’s note that cognition is interactive and embodied, this is a straw-man that has been re-hashed repeatedly by those who want to try and “take down” cognition for decades now. Its consistent rehashing causes many cognitive scientists just to roll their eyes at the critique.
