Filter and transmission models

What it is

The proposal that the brain does not generate consciousness but transmits, filters or constrains it — that mind is in some sense prior to the brain, and the brain's function is to limit and localise it rather than to produce it.

The standard analogy, and its standard weakness: a radio does not create the broadcast, and smashing the radio stops the music without proving the music was inside it. The weakness is that a radio is an artefact designed by someone to receive a signal we can independently detect, and neither condition holds for the brain.

Where it came from

William James gave the most careful statement in Human Immortality (1898), distinguishing the brain's productive function from its transmissive or permissive one, and arguing that the evidence of neurology — that damaging brains damages minds — is equally consistent with both, so that materialism's claim to have settled the question by that evidence is unearned. He did not assert the transmissive view; he asserted that the argument against it had not been made.

Henri Bergson developed a version in Matter and Memory (1896), with the brain as an organ of action rather than of representation, selecting from a wider field of memory and awareness what is useful for present conduct.

Frederic Myers's subliminal self is the same family of idea worked out empirically.

Aldous Huxley popularised it as the reducing valve in The Doors of Perception (1954), borrowing from the Cambridge philosopher C.D. Broad: that each person is at every moment capable of remembering everything and perceiving everything, and the nervous system's function is to protect us from being overwhelmed by knowledge that is mostly useless.

How it is used

As the implicit framework for almost every claim in this section. Near-death experiences during minimal brain activity, terminal lucidity in destroyed tissue, psychedelic experience as opening rather than distortion, mediumship, survival — all of them make sense on a filter model and none of them does on a straightforward generative one. The model is rarely stated and almost always assumed.

It has also had a recent scientific outing: the entropic brain hypothesis and imaging findings that psychedelics reduce activity in the default mode network while producing a subjective expansion of experience were greeted as vindication of the reducing valve. That interpretation goes well beyond what the data support — disrupting an integrative network can perfectly well produce richer experience within a generative model — but the finding is genuine and the discussion is live.

What we can and cannot say

We can say the logical point is sound. Correlation between brain states and mental states is exactly what both models predict. Neuroscience has established the correlation in extraordinary detail and has not, by that alone, distinguished production from transmission. Anyone who says the brain scans settle it has not understood the objection.

We can say the model has almost no positive evidence. That an argument against it fails is not an argument for it. To be more than a logical possibility it needs to predict something the generative model does not, and then have that prediction confirmed. It has not done this.

We can say it is not currently falsifiable, and that this is its central weakness. Whatever neuroscience finds — this region does that, this lesion abolishes this capacity — can be absorbed as a fact about the filter. A model that cannot fail is not doing scientific work, whatever else it may be doing philosophically.

We can say it faces a hard specificity problem. Damage to a small area of the left hemisphere does not merely dim consciousness; it removes the ability to name objects while leaving everything else intact. Damage elsewhere removes the ability to recognise faces, or to form new memories, or to feel fear. A filter that can be selectively broken in ways that map precisely onto specific cognitive functions is doing so much of the work that it has become indistinguishable from a producer. Explaining that away is the model's real burden, and it has not been carried.

We can say it has respectable philosophical company. It sits close to idealism, to some forms of panpsychism, and to the property-dualist end of the mind-body problem, and it is a live option in contemporary philosophy of mind rather than a fringe one. It is also, essentially, what Advaita Vedānta and several Buddhist schools have been saying for a very long time in a different vocabulary.

We cannot say the filter model is true or false. What we can say is that it is where every claim in this section eventually has to go, that it has been available since 1898, and that in a hundred and twenty-five years nobody has found a way to test it. Knowing that is the most useful thing this section can leave a reader with.

Further reading

• William James, Human Immortality: Two Supposed Objections to the Doctrine (1898) — short.
• Kelly, Kelly et al., Irreducible Mind (2007) — the fullest modern case, and openly partisan.
• Carhart-Harris and Friston on the entropic brain, for the current scientific version of the argument.