Myths about the Brain

False facts about the brain and neural code are often repeated and  tought to be correct. Unfortunately,  journals with high impact factor are among the ones which  have frequently reinforced  these misconceptions.

1. Myth: Action potentials are digital signals

Hopfield JJ. 1995 Pattern recognition computation using action potential timing for stimulus representation. Nature. Jul 6;376(6535):33-6.
Gerstner, W, 2007, A framework for spiking neuron models: The spike response model, Handbook of Biological Physics, Volume 4, 2001, Pages 469–516, 

 

Truth:   Action potentials are fast (<1 millisecond) all or none phenomena,  however they are not  digital events http://neuroelectrodynamics.blogspot.com/p/spike-directivity.html. The spatial modulation of APs, the recorded shape carry meaningful information.   Our convenience (digital action potentials) is our MISTAKE that has separated the generation of  electrical patterns from  neural computation and molecular biology.

Note: Experimental data show clear evidence  of spatial, meaningful  modulation of action potentials  see  http://www.sciencedirect.com/science/article/pii/S0165027012001021?v=s5 and references at bottom of page http://neuroelectrodynamics.blogspot.com/p/spike-directivity.html , 

Dorian Aur,  A Comparative Analysis of Integrating Visual Information  in Local Neuronal  Ensembles, Journal of Neuroscience Methods, 2012, Volume 207, Issue 1, 30, Pages 23–30,  http://www.sciencedirect.com/science/article/pii/S0165027012001021

2.Myth: The neural code is hidden in temporal patterns

 

Shadlen M.N., Newsome W.T. Noise, neural codes and cortical organization

      (1994) Current Opinion in Neurobiology, 4 (4), pp. 569-579.

W Gerstner, R Kempter, JL Van Hemmen, 1996, : A neuronal learning rule for sub-millisecond temporal coding, NATURE  Volume: 383   Issue: 6595   Pages: 76-78

Friedrich RW; Laurent G, 2001, Dynamic optimization of odor representations by slow temporal patterning of mitral cell activity,  SCIENCE,   Volume: 291   Issue: 5505   Pages: 889-894 
Jones LM; Depireux DA; Simons DJ; 2004, Robust temporal coding in the trigeminal system, SCIENCE   Volume: 304   Issue: 5679   Pages: 1986-1989
Kayser Christoph; Montemurro Marcelo A.; Logothetis Nikos K.; 2009   Spike-Phase Coding Boosts and Stabilizes Information Carried by Spatial and Temporal Spike Patterns, NEURON   Volume: 61   Issue: 4   Pages: 597-608  
    

Truth:  They forgot to mention that recorded temporal patterns were in fact generated by a  series of complex electrical events. Since information is intracellularly processed (inside neurons) then temporal  patterns carry little information  http://neuroelectrodynamics.blogspot.com/p/temporal-coding-realistic-model-of.html. The required fragments of information are distributed within macromolecular structures in  many neurons. Therefore, the sub-millisecond temporal occurrence of APs  is required  to electrically integrate in the brain information read from molecular structure during spiking activity (action potentials, synaptic activity) http://neuroelectrodynamics.blogspot.com/p/from-spike-timing-dogma-to.html .  The ‘synchronous’  activation provides  the  ‘short term memory’ access needed to recognize objects, understand concepts...  http://neuroelectrodynamics.blogspot.com/p/cognition-and-consciousness.html. 

3.Myth: There is ‘noise’ in  temporal  patterns

London M, Roth A, Beeren L, Hausser M, and Latham PE (2010) Sensitivity to
perturbations in vivo implies high noise and suggests rate coding in cortex.
Nature 466: 123–127.

Truth:  The presence of “noise” in neuronal activity   is another flaw  generated  by a false  hypothesis  --- digital, stereotyped spike. Indeed, if  solely temporal patterns are considered then neuronal activity  may appear random since  information (semantics) is elsewhere  http://neuroelectrodynamics.blogspot.com/p/spike-directivity.html
The  activity in the brain is less random than expected.  Periods with increased randomness are rare events  in the brain- persistent  chaotic dynamics  occurs right  before seizure generation  http://www.ncbi.nlm.nih.gov/pubmed/21679727 .  In addition, chaotic behavior is not a completely random phenomenon.

4.Myth: Our memories are preferentially stored in the  hippocampus

Jensen, O. and Lisman, J.E. 2005, Hippocampal sequence-encoding driven by a cortical multi-item working memory buffer. Trends in Neuroscience, 26, 696-705.

Squire LR, Memory and the Hippocampus - A Synthesis From Findings With Rats, Monkeys and Humans Psychological Review  Volume: 99   Issue: 2   Pages: 195-231

Truth:  All neurons can store different fragments of information (memory) in their molecular structure (proteins) not only neurons from hippocampus http://neuroelectrodynamics.blogspot.com/p/computing-by-interaction.html. Information is integrated  during spike generation (APs, synaptic activities)  http://neuroelectrodynamics.blogspot.com/p/cognition-and-consciousness.html  

5.Myth: The memories can be ‘read’ by knowing the connectivity between neurons

Sebastian Seung 2012 Connectome: How the Brain's Wiring Makes Us Who We Are, New York: Houghton Mifflin Harcout.

Seung H. S.,  2009, Reading the Book of Memory: Sparse Sampling versus Dense Mapping of Connectomes,    Neuron, Volume 62, Issue 1, , Pages 17–29

Truth:  The entire attempt to read memories is based on a  false  hypothesis that memories are solely written in the connections between neurons . This  assumption  excludes the most important part -- the neuron where information  is stored and processed http://neuroelectrodynamics.blogspot.com/p/computing-by-interaction.html .  This information stored within  molecular structure is 'read'  during AP propagation which shows fast meaningful dynamics on the submillisecond time scale http://neuroelectrodynamics.blogspot.com/p/spike-directivity.html.  The resulting  ‘connectivity’ follows AP propagation ( information processing) which can change with every generated spike. In addition, the function of neurons can be totally reshaped during learning (3-4 days to one week). The effect is  a local  change in interactions between neurons (connectivity). Therefore,  the entire gimmick to "read" out memories at the brain scale based on ‘connectivity’ represents a  naïve,  obsolete idea.

(i)                 We are who we are since simple cells have evolved into complex neurons, are densely packed in the brain to allow  electrical integration of information during spiking activity.

(ii)                We are unique in the way we operate among animal species. We are who we are since our  brain  architecture and  molecular structure embedded in neurons  reflect  genetically encoded information  from DNA segments, a result of evolutionary processes.

(iii)              We are who we are since  a continuous process of  interaction with our peers has shaped our behavior from the first day we were born

6.Myth: Memory is stored in synaptic connections between neurons 

Bliss TVP; Collingridge GL 1993, Experimental data A Synaptic Model Of Memory - Long-Term Potentiation In The Hippocampus,  NATURE  Volume: 361   Issue: 6407   Pages: 31-39

Truth: The absolute number of synapses per neuron is reached by age 1 year and  strongly decreases during the preschool years. The dynamics of AP generation and spatial propagation makes synaptic connectivity between neurons  highly variable. While information stored in the brain increases with time, the  number of synapses reduces  with age. Our memories are stored within molecular structure (proteins) distributed in neurons http://neuroelectrodynamics.blogspot.com/p/from-spike-timing-dogma-to.html   in addition to molecular layer embedded  in synapses http://neuroelectrodynamics.blogspot.com/p/spike-directivity.html

7.Myth:  This neuron responds to  a specific  concept -  Concept cells

Quiroga, R.Q., Reddy, L., Kreiman, G., Koch, C., Fried, I.  (2005) Invariant visual representation by single neurons in the human brain   Nature 435 (7045), pp. 1102-1107
http://www.scientificamerican.com/article.cfm?id=being-john-malkovich (Christof Koch)

Truth:  The firing rate analysis is not sensitive enough to detect which information is intracellularly processed  http://neuroelectrodynamics.blogspot.com/p/spike-directivity.html . Concept cells are grandmother cells http://neuroelectrodynamics.blogspot.com/p/concept-cells.html . The existence of grandmother behavior  was  previously strongly rejected  by Jerome Lettvin

http://jerome.lettvin.info/lettvin/Jerome/WhatTheFrogsEyeTellsTheFrogsBrain.pdf

8.Myth: Artificial neural networks (ANNs)   describe computations performed by  biological neurons 

Mahowald M; Douglas R, A Silicon Neuron, 1991, Nature,   Volume: 354   Issue: 6354   Pages: 515-518

Truth:  The claim that  artificial neurons are simplified models of their biological counterparts is untrue. Artificial neurons are abstract mathematical models (spiking neurons, weight type connection)  they represent  attempts to  build  optimal, adaptive systems, unfortunately they  do not describe  biophysical properties  of neurons http://neuroelectrodynamics.blogspot.com/p/from-spike-timing-dogma-to.html or  computations performed by biological  neurons

http://neuroelectrodynamics.blogspot.com/p/computing-by-interaction.html 

9. Myth: Axons, dendrites  function as electrical cables

Mahowald M; Douglas R, A Silicon Neuron, 1991, Nature,   Volume: 354   Issue: 6354   Pages: 515-518

Truth: The process of computation in neurons is not just  information transmission through cables http://neuroelectrodynamics.blogspot.com/p/from-spike-timing-dogma-to.html ,  it  involves a complex process of  interaction http://neuroelectrodynamics.blogspot.com/p/computing-by-interaction.html 

 Electrical conduction  (wiring) does not model any  transfer of information between molecular structures and electric flow which  occurs in neurons within dendrites, axons during AP propagation.

10. Myth:  There is  a unique flow of information from genes by transcription translation  to protein structures

  Crick F. (1970) Central dogma of molecular biology, Nature, 227 (5258), pp. 561-563.

Truth: The presence  of  posttranslational modifications, epigenetic effects of chemical signal at several levels (e.g. chemical modifications to DNA) regulatory interactions that operate at gene and protein level disprove the central dogma. In addition the effect of electrical interactions indicates a more complex mechanism  for information flow than earlier proposed and contradicts the main dogma  which limits genetic-causal explanations.
 

11. Myth:  Neuroscientists Capture Real-Time Brain Dynamics Via .....

Lin LN; Osan R; Shoham S; 2005, Identification of network-level coding units for real-time representation of episodic experiences in the hippocampus, PNAS, Volume: 102   Issue: 17   Pages: 6125-6130 

 

Truth:  Information processing and communication are extremely dynamic phenomena in the brain. Within neurons relevant information is  processed on the sub-millisecond time scale.  Many technique cannot  capture these  fast  dynamics. In addition statistical methods and averaging technique over several repetitions of stimuli or behaviors hide relevant, specific details.  Therefore, in spite  of  many claims only  few techniques  are able to capture meaningful  dynamics  related  to information processing or memory encoding  http://neuroelectrodynamics.blogspot.com/p/spike-directivity.html. In addition the role of neurons can significantly change with learning during several repetitions of stimuli or behaviors over few days of training.

Direct Consequence: Unreliable outcome regarding brain dynamics

12. Myth: Brains  are essentially connectionist models of computation - fundamentally prediction machines - perform Bayesian inference 

Knill DC; Pouget A, 2004 The Bayesian brain: the role of uncertainty in neural coding and computation, Trends in Neurosciences  Volume: 27   Issue: 12   Pages: 712-719   

Truth: Connectionist models represent an attempt to provide the  simplest  model of mind/ brain. From connectivity in  proteins or between genes to the association of planets and days of the week, all types of  interactions can be approximated by weight type connections. Therefore, the  idea of weight type connectivity presents a general,  non-specific model . However, the brain has a specific model of computation by physical (electrical) interaction mediated by molecular changes in neurotransmitters levels  http://neuroelectrodynamics.blogspot.com/p/computing-by-interaction.html. The continuous (non-Turing) model of computation by electric interaction intrinsically exhibits many features  such as parallelism, fuzziness, fractality  in addition to predictive or Bayesian appearance.   Digital principles borrowed from engineering or machine learning (e.g. prediction error) reflect just tiny parts of multiple computational features expressed by biological neurons http://neuroelectrodynamics.blogspot.com/p/computing-by-interaction.html 

13. Myth: The brain shows a  strict hierarchy of information processing 

Bodegard A; Geyer S; Grefkes C; Hierarchical processing of tactile shape in the human brain, NEURON  Volume: 31 , Pages: 317-328  

Truth:  The idea of hierarchy in connectionist models has represented an attempt to model anatomical (hierarchical) organization. Currently, little experimental data supports the notion of a strict hierarchy in information processing. The presence of various forms of computation at the sub-cellular level http://neuroelectrodynamics.blogspot.com/p/computing-by-interaction.html   and continuous electrical integration of information within neurons http://neuroelectrodynamics.blogspot.com/p/from-spike-timing-dogma-to.html do not highlight a strict hierarchy. Related to a specific task or behavior the simultaneous activation of neurons in the brain does not seem to follow a strict anatomical (hierarchical) propagation. Since fragments of information are distributed in various neurons then  sequential/ parallel   activation of specific cells  is required to integrate information needed for perception or action.  Either they spike or not all neurons process received information . The brain in all its entirety  is the computing machine that exploits  parallel distributed processing within many  neurons. A strict hierarchy of information processing  represents a non-specific reductionist  model of developed interactions http://neuroelectrodynamics.blogspot.com/p/cognition-and-consciousness.html

14. Myth: Always mathematical  (statistical) models   help us  to understand complex processes that occur in the brain

 Knill DC; Pouget A, 2004 The Bayesian brain: the role of uncertainty in neural coding and computation, Trends in Neurosciences  Volume: 27   Issue: 12   Pages: 712-719   

Truth: False hypotheses about a phenomenon may strongly limit the explanatory power of any mathematical model. Theoretical constructs regarding  Bayes theory, nonlinear dynamics  ...  HAVE NO REAL VALUE  in understanding information processing in neurons, in the brain if  they are attached to a false hypothesis (DIGITAL spike) http://neuroelectrodynamics.blogspot.com/p/temporal-coding-realistic-model-of.html

15. Myth: Learning can be mainly described by changes in synapses 

Bliss TVP; Collingridge GL, 1993, A Synaptic Model Of Memory - Long-Term Potentiation In The Hippocampus,NATURE  Volume: 361   Issue: 6407   Pages: 31-39  

 

 Truth: Learning is a result of reorganization at a molecular level that occurs in neurons in addition to synaptic changes. The entire learning process is deeply rooted within specific molecular and genetic mechanisms of protein synthesis that include gene selection, regulation and expression related to DNA-RNA transcription, mediated by electrical interactions, neurotransmitters and hormone levels.   The effect of  above  changes  is partially reflected in well observed re-organization of spike patterns measured by common electrophysiological recording techniques and interpreted as forms of learning (habituation, sensitization, classical conditioning).

 

Temporal patterns indicate solely  WHEN electrical events occur which  PARTIALLY characterize  any electric event, does not tell WHAT information was electrically communicated. Few have  realized that temporal patterns were in fact  the moments when meaningful electrical patterns  occur within the neuron http://neuroelectrodynamics.blogspot.com/p/spike-directivity.html . 

 The entire framework, the  analysis of various time scales ( firing rate, ISI, spike timing dependent plasticity ) represent just a   tiny part of information needed to characterize electrical events  - that's the fundamental problem with temporal coding. Therefore, these myths about the brain represent  a gentle introduction to NeuroElectroDynamics

 Max Planck  ---- “A new scientific truth does not triumph by convincing its opponents and making them see the light......"

 David Deutsch -----" There is indeed an objective difference between a false explanation and a true one, between chronic failure to solve a problem and solving it, and also between wrong and right, ugly and beautiful, ......."

NeuroElectroDynamics or NED is the study of  the dynamics and interaction of electrical charges in the brain [1]. The word neuroelectrodynamics is derived from neuro- meaning neurons, electro- electric field and -dynamics meaning movement.

The main idea of NED is that under the influence of electric fields, charges that interact perform computations and are capable to read, write and store information in their spatial distribution at molecular level within active neurons. The universal physical laws from classical mechanics, thermodynamics to quantum theory can be applied to generate a consistent mathematical model of brain computation.

The fundamental claim of NED is that temporal observables associated with neural coding ( temporal coding, spike timing occurrence, spike-timing-dependent plasticity, ) are epiphenomena determined by the dynamics and interaction of electric charges modulated by molecular changes in neurotransmitters levels, regulatory mechanisms of gene expression from DNA to proteins synthesis.

NED  highlights a specific form of computation by interaction which is a general physical model of computation extensively present in nature [6].

A spontaneous generation of action potentials and synaptic activities is needed to maintain physical interaction. Meaningful information encoded (written) within neurons and synapses at a molecular level  can be   transmitted synaptically and non-synaptically  during action potential propagation [1][5].

During these events, the required information is exchanged between molecular structures (proteins), which store fragments of information, and the generated electric flux, which carries and integrates meaningful information in the brain [5].

History

Early work started with a fundamental electrophysiological observation. Contrary to common belief, action potentials generated by the same neuron are not alike, they display changes in electrical patterns not just temporal variability.

Every recorded action potential can be characterized by a new measure, spike directivity that describes electrical activity in a biological neuron [2]. Significant changes in spike directivity are correlated with changes in behavior [3]. Since information is carried by electric charges [4], then their dynamics and interaction characterize complex computational processes in the brain.

References

[1] Aur D., Jog, MS., 2010 Neuroelectrodynamics: Understanding the brain language, IOS Press, 2010.

[2] Aur D., Connolly C.I., and Jog M.S., 2005 Computing spike directivity with tetrodes. J. Neurosci. Vol. 149, Issue 1, pp. 57–63.

[3] Aur D., Jog, M.S., 2007 Reading the Neural Code: What do Spikes Mean for Behavior? Nature Precedings, http://dx.doi.org/10.1038/npre.2007.61.1

[4] Aur D., Connolly C.I. and Jog M.S., 2006 Computing Information in Neuronal Spikes, Neural Processing Letters, 23:183-199.

[5] Aur, D. 2012,  A comparative analysis of integrating visual information in local neuronal ensembles. Journal of neuroscience methods, 207(1), 23-30.  http://www.ncbi.nlm.nih.gov/pubmed/22480985

[6] Aur D, Jog MS, Poznanski, R, Computing by physical interaction in neurons, Journal of integrative Neuroscience, vol. 10, Issue: 4, 2011, pp. 413-422http://www.ncbi.nlm.nih.gov/pubmed/22262533