Quantum neural network
Recurrent Quantum Neural Network & its Applications
Laxmidhar Behera, Indrani Kar & Avshalom Elitzur
Indian Institute of Technology, & Bar-Ilan Institute, Israel
In: Tuszynski, J. Ed. The Emerging Physics of Consciousness Springer ISBN-13 978-3-540-23890-4
This chapter sidesteps the whole question of whether quantum activity exists in the brain, and goes straight to discussing the need for some sort of quantum effect to explain some of the things that the brain is able to do, particularly the saccade movements of the eyes. The model proposed here suggests that there is a quantum process that effects the average behaviour of a neural lattice. The authors remind us that some of the tasks humans perform with ease, such as more difficult forms of pattern recognition are still beyond the capacity of super computers. The authors look for a quantum mechanical explanation of the abilities of orgnisms in this respect, and the chapter discusses human eye movement from this respect.
Simulations of brains produce some interesting insights in this respect. When the eye tracks a target, a related wave packet moves not in a continous classical manner but in a discrete quantum manner, which is considered similar to the ‘jumps’ and ‘rests’ involved in saccidic eye movement. Eye movements relative to static scenes are not continous but involve discrete ‘jumps’. If the information is new or difficult to interpret there are more erratic pauses and even a process of flip back and forth between different images. This again is similar to the movement of quantum wave packets in simulations. Eye tracking experiments show that smooth eye movements contain errors, which are corrected by saccades that bring the eye back to the required position. After one or two quick saccades the eye usually becomes adjusted to the target. Saccadic and smooth movements are seen to combine to keep track of the visual target.
The authors suggests a view of neural information processing where quantum processing effects a neural lattice with spatial structure. This is known as a Recurrent Quantum Neural Network model or (RQNN). This is distinct from other models. This quantum based model is said to be very succesful in explaining the actual nature of eye movements, and to be 1,000x more accurate than conventional models. The authors hope that their work will encourage researchers to study the brain from a quantum perspective. Microtubules are mentioned as the most plausible location for quantum activity in the brain, with electrons inside hydrophobic pockets localised towards either to the A or B end of the tubulin dimer and capable of being in a superposition of these states. Apart from this, the authors simply describe their model as idealised, and decide to remain silent on the question of the actual physical structure of quantum information processing or the problem of decoherence.
Behera, L. & Sundaram, B. (2004) Proceedings, International Conference on Intelligent sensors
Behera, L. et al (1996) IEE Trans Neural Networks, 7 (6) pp. 1401-1414
Behera, L. et al (1998) IEE Proceedings Control Theory and Applications, 145 (2) pp. 134-40
Behrman, E. et al (2002) Physical Review Letters
Behrman, E. et al (2000) Information Sciences, 128, (3-4) pp. 257-69
Davydov, A. (1982) Biology and Quantum Mechanics Pergamon Press
Gupta, S. & Zia, R. (2001) Journal of Computer and System Sciences, 63 (3) pp. 355-383
Hagan, S., Hameroff, S. & Tuszynski, J. (2002) Physical Review E, 65, pp. 061901
Atmanspacher, H. (2004) Discrete Dynamics, 8, pp. 51-73
Mershin, A., Nanapoulos, D. & Skoulakis, E. (1999) Proceedings of the Academy of Athens, 74, pp. 148-79
Tuszynski, J., Hamerof, S., Sataric, M. et al (1995) Journal of Theoretical Biology, 174, pp. 371-380