Neural Substitution Fallacy
It can be argued that this neuro substitution argument is the primary reason “functionalism”
has become and remains the leading consensus theory.
is a supporting sub camp of “Representional Qualia Theory”
which defines consciousness to be “computationally bound elemental physical qualities like redness and greenness.” While at first glance it appears that the neuro-substitution argument includes redness and greenness functionality, the substitution is always done on a system that does not include the required “binding mechanism.” The absence of a binding system enables the sleight-of hand trick that does not include redness and greenness functionality in a sufficient capacity.
To illustrate the fallacy, let’s do a neuro substitution on a contrived, overly simplistic theory that is not qualia blind (does include these necessary qualia functions) to illustrate the fallacy. Since Molecular Materialism
is a simple theory, let’s start with a variant of that. Let’s assume the neurotransmitter glutamate, reacting in a synapse, has a redness quality, and glycine has a greeness quality. Let’s say there is a single neuron representing each pixel of visual knowledge of which we are consciously aware. If these neurons dump glutamate into the synapse, that pixel of knowledge has the physical glutamate redness quality, the same for when that pixel neuron switches to dumping glycine greenness when that single pixel on the surface of the strawberry turns green. In other words, for each point on the surface of a red strawberry, there is a neuron firing with redness glutamate, and for each point on the green leaves, there is a neuron firing with greenness glycine when we are consciously aware of a strawberry.
Let’s say there is a very large single downstream neuron detecting the glutamate or glycine from all these upstream pixel neurons that perform the necessary computational binding to compose a complete composite qualitative conscious awareness of a strawberry. This binding neuron would necessarily be a glutamate (redness function) detector. Anything other than redness functionality being presented to this binding neuron must produce a “not redness” output, otherwise the required ability to reliably detect glutamate functionality would not be preserved.
Let’s simply use a binary 1 to simulate redness, and 0 to represent greeness. In parallel, let's do a redness / greenness substitution to further illustrate the problem. In this example, the first pixel neuron firing with redness in the middle of a patch of redness on the strawberry is what is replaced. The simulated neuron is now outputting a binary 1. The inverted version is dumping greenness glycine instead of redness glutamate. In order for the binding neuron to say the particular patch including this pixel was entirely red, you’d need to provide an interface system, as described in the argument, that would present glutamate to the real binding neuron when the simulated neuron produced a 1. In parallel, the inverted neuron is enabling greenness to be mapped back to redness.
Let’s progress to the point where about half of the real pixel neurons are replaced with simulated ones. At this point, we want to be able to switch between the real binding neuron and the simulated binding neuron. With the switch in the real binding neuron position, all the simulated neurons’ 1s need to be mapped back to redness to feed the binding neuron synapses. Then, when you switch to the simulated binding neuron, this mapping system from the simulated pixels is no longer necessary, as the artificial binding neurons just take 1. Since half of the pixels are still real, they are dumping glutamate, which now needs to be mapped back to 1, which the simulated binding neuron requires. In other words, when the switch is in the simulated position, everything being presented to the binding neuron is either all 1s, or all green. When it is in the real binding neuron position, everything being presented is glutamate redness. This extreme amount of change being done with one atomic switching action completely removes any knowledge from the system which could inform it of the significant internal qualitative functional change that results in the seemingly consistent output: “that is all red”. This is the sleight-of-hand.
The provided binding mechanism must also enable the “strongest form of effing the ineffable” (See referenced material in “Representional Qualia Theory”
). In other words, you could use an Avatar-like neural ponytail
to connect to the neural substituted system. A neural ponytail
would enable someone to directly experience all of the experiences, not just half. If the necessary binding system wasn’t enabling you to detect whether the target neuro substituted system was using redness or greenness, or 1s to represent red things, it wouldn’t be functioning correctly.
Canonizer always stresses how falsifiability significantly improves the quality of camps. Towards this end, we admit that if anyone can provide a description of a theory that meets the above necessary definitions of consciousness, and then describe a neural substitution on that system in a way that there is no such sleight-of-hand (without doing violence to Occam’s Razor
), the supporters of this camp would consider this camp to be falsified and abandon it. We would then leave it up to the experimentalists to determine which of these theories, if any, is THE ONE true theory that can’t be falsified.