Holonomic brain theory
[4][5][6][7][8] Gabor, Pribram and others noted the similarities between these brain processes and the storage of information in a hologram, which can also be analyzed with a Fourier transform.
[13] One of Gabor's colleagues, Pieter Jacobus Van Heerden, also developed a related holographic mathematical memory model in 1963.
[14][15][16] This model contained the key aspect of non-locality, which became important years later when, in 1967, experiments by both Braitenberg and Kirschfield showed that exact localization of memory in the brain was false.
On the other hand, Pribram removed large areas of cortex, leading to multiple serious deficits in memory and cognitive function.
[17] Several years later an article by neurophysiologist John Eccles described how a wave could be generated at the branching ends of pre-synaptic axons.
[18] After studying the work of Eccles and that of Leith,[17] Pribram put forward the hypothesis that memory might take the form of interference patterns that resemble laser-produced holograms.
[17] Pribram was further encouraged in this line of speculation by the fact that neurophysiologists Russell and Karen DeValois[22] together established "the spatial frequency encoding displayed by cells of the visual cortex was best described as a Fourier transform of the input pattern.
In each smaller individual location within the entire area it is possible to access every channel, similar to how the entirety of the information of a hologram is contained within a part.
The beam always contains all the information of the object, and when conjugated by a lens of a camera or the eyeball, produces the same full three-dimensional image.
Holographic models can also demonstrate associative memory, store complex connections between different concepts, and resemble forgetting through "lossy storage".
[5] Many synaptic locations are functionally bipolar, meaning they can both send and receive impulses from each neuron, distributing input and output over the entire group of dendrites.
[4] Pribram posits that the length of the delay of an input signal in the dendritic arbor before it travels down the axon is related to mental awareness.
A study by David Alkon showed that after unconscious Pavlovian conditioning there was a proportionally greater reduction in the volume of the dendritic arbor, akin to synaptic elimination when experience increases the automaticity of an action.
It had been thought that binding only occurred when there was no phase lead or lag present, but a study by Saul and Humphrey found that cells in the lateral geniculate nucleus do in fact produce these.
[11][31] These may play a role in cell communication and certain brain processes including sleep, but further studies are needed to strengthen current ones.
Though not directly related to the holonomic model, they continue to move beyond approaches based solely in classic brain theory.
The Gabor model did not explain how the brain could use Fourier analysis on incoming signals or how it would deal with the low signal-noise ratio in reconstructed memories.
Longuet-Higgin's correlograph model built on the idea that any system could perform the same functions as a Fourier holograph if it could correlate pairs of patterns.
[3] Like a hologram, a discrete correlograph can recognize displaced patterns and store information in a parallel and non-local way so it usually will not be destroyed by localized damage.
[34] One property of the associative net that makes it attractive as a neural model is that good retrieval can be obtained even when some of the storage elements are damaged or when some of the components of the address are incorrect.
