Complexity in Searching for the Neural Code

FEATURE CONNECTED iStock_000023922412SmallNeuroscience searches for an algorithm known as the neural code. Such a code translates the firing of neurons into mental events including thoughts and emotions. Current large scale attempts to “map” the brain are based on this concept. This post will address some of the many, extremely difficult, problems with this approach.

A previous post on the connectome already outlined some issues with the mapping approach: a changing moment by moment neuroplasticity with new and pruned synapses each day, each placed in new spatial relationships on the dendrite; large numbers of different types of neurons in different brain regions; extremely complex synaptic extracellular matrix; post synaptic densities that are different in different regions (each including thousands of large complex proteins which are different in different regions). Another major issue noted was that neurons communicate in more ways than axon signaling, for example, synchronous electromagnetic brain waves and ephaptic sideways signaling from along the axon. Another range of problems involves the fact that neurons themselves are vast with many different ways to analyze its own information before any communication with other neurons.

But, there are many other challenges in this approach.

Synchrony Neural Code

dendritic treeThe temporal spatial qualities of the axon firing present its own set of problems. The topological placement of the synapses on the dendrite appears to be relevant, and they constantly change through neuroplasticity and budding dendrite spines. The temporal issues include the fact that pulses are at times rhythmic and other times not; they have different amplitudes.

A search for the rhythmic code has been called the synchrony code. Some specific firing rates have been correlated with mathematical curves such as visual orientation, motion selection, or target location when reaching for an object. Neurons are exquisitely sensitive to any changes in the temporal patterns. While synchronous coupling has been observed in different regions of the brain, synchrony code refers instead to inherent precise timing of multiple closely related neurons at the same time. Research shows that unless the events occur within 10 milliseconds of each other, then the synchrony is lost. In the thalamus for example, bursting has been observed, which appears to be a signal for a circuit to pay attention to the next sequences. Then, there is synchronous firing of many individual neurons during this “window”. This implies that the synchrony can create “windows of opportunity” for other signals.

Through study of this synchrony code, another very difficult problem for research has been observed. It appears that among a region of neurons, specific neurons are used in different patterns with a variety of neuron subgroups. It is not at all clear how a neuron could know to participate in one circuit at one moment and a completely different group at the next moment.

Interestingly, recent research on hearing implants shows that the brain will respond with new types of codes for the new unusual circumstances. This might mean that the brain invents codes for particular types of stimuli and then somehow keeps track of this.

More Complications – Changing Neurotransmitters, Extra cellular Matrix and Astrocytes

Most neurons have specific neurotransmitters that they use, but for particular circumstances they can change and send new neurotransmitters. A recent study showed switches in a neuron that completely altered the function of the neuron in winter and in summer.

untitled211316472995722The previous post on astrocytes discussed another enormous problem with considering mapping the neurons as switches. Current research shows that it is astrocytes (not the neurons), creating the matrix for much of the volume of the brain, which determine all of the important developmental stages and functions of the synapse. In fact, without the constant presence and stimulation of factors from astrocytes, synapses would not occur.

Astrocytes also control the blood flow with end feet that constrict and dilate the blood vessels, and determine the fRMI signal. Please see post “Astrocytes Control Synapse Function”, for a detailed summary of the critical astrocyte functions. Most difficult, perhaps, of all issues for the neuronal map, is that the astrocytes have their own calcium based action potential that sends critical information throughout the brain.

Very recently, an entirely different issue is raised about “whole brain” activity.

Whole Brain Activity and Recordings

MRI can observe the blood flow of the entire brain, but only in the time scale of seconds. Neuronal or astrocyte events occurring throughout the brain in a range of milliseconds might have no visible impact on the blood flow, but can be very significant. Fleeting connections of small numbers of neurons will not be seen. The fMRI only shows the average of these thousands of millisecond events over the span of a second. 

A recent study took fMRI for a half hour while people watched a series of videos of scenes from everyday life. The videos contained plants, buildings, animals, people and other objects. The total brain was divided into a large grid on fMRI where each small brain section was correlated with the content of the video by type of object. People were told to either look at the people or vehicles.

When the people looked for humans, the entire brain was active looking for the people in the videos. When asked to focus on vehicles, the brains were also totally active in a different way, and also were able to see many other objects as well as vehicles.

people molecular structureWhat was very striking in this experiment is that by changing attention the entire brain changed, not one region. The focus of the mind changed the entire brain. The only way to understand the total brain activity was to know what the person was thinking about.

Another study looked at the brain waves and oscillations of the entire brain during activities such as tapping a rhythm with fingers. The study found rolling activity over the entire brain for activities that had previously been thought to be located in one region. This rolling wave was not random but had a specific pattern for each different activity. In fact, repeating an action shows different patterns since there is learning from the first tap – conscious activity is eventually trained in the habit memory regions to become an unconscious behavior (e.g., driving a car without paying attention, suddenly driving a bicycle after not driving twenty years).  In children, uniquely, the waves went from the back of the brain to the front area.

Researchers are now looking into mathematical ideas to take into account this wide-ranging activity in the whole brain at once. One study looked at correlating a backbone brain circuit, connecting all physical hubs, with another circuit related to blood flow changes in specific regions. This phrenologywork attempts to understand how wide ranging activity can occur with synchronized activity at a distance.

The point is that the deluge of studies that overly interpret brain regions has become a new form of “phrenology.”

Distributed Memory

Recent research shows that there is no single memory region (much like the recent discovery of multi sensory connections in all previously thought individual sensory regions). It is shown that there are synchronous brain waves from different neuronal regions, which form an electrical couple connecting the two regions.

Strikingly, recent research shows memory related to time and to space have different coupling connecting specific small regions of pre frontal, parietal and cortex near the hippocampus.

Previously it was thought that explicit memory circuits (a scene with buildings, location, color, sounds, and feelings) connected the cortex and hippocampus, with implicit memory using multiple different parallel circuits. However, each of the circuits also connects with many other brain regions that relate to the specific type of sensory, motor and emotional systems involved with the specific memory. The question has been how do these changing very complex connections all over the brain send enough information to organize the information for use in an ordinary moment.

Memory is not a single process but exists in broad circuits around the brain. There is really very little known about how a single memory is formed. The recent study focused on time and place, two key elements of an explicit memory of a scene – where and when it happened. Remarkably, the two critical aspects of explicit memory couple the specific involved regions with two different specific frequencies of synchronous brain waves, called a “spectral fingerprint” or “footprint.”

This implies that many of the brain circuit communications are not occurring by wired axon communications, but by wireless communication within the system (see my post in Scientific American Mind, Wired and Wireless Brain.). Somehow, the specific frequencies of the oscillations facilitate the electric wire spiking for information transfer.

‘Space’ Information Oscillates at 1 to 4 Hz – ‘Time’ Information 7 to 10 Hz

97-03-27-megajas_mega1In these studies, increased low frequency coherence was noted for both ‘time’ and ‘space’ in the range of 1 to 10 cycles per second. The region around the hippocampus was the critical region for both space and time. But, the band of 1 to 4 related to information about space, and 7 to 10 about time. Both circuits included different regions   with ‘space’ content connecting with pre frontal and precuneus and ‘time’ content prefrontal and inferior parietal.

freq-wavelengthThis research was painstaking and involved knowledge of exactly where to place electrodes in live human subjects. The nature of the wildly distributed brain regions involved in real memory is still far too complex. It is known that a much higher frequency gamma oscillation (70 to 250 cycles per second) is very relevant to specific axon spiking in language and motor tasks.

One important question is the code for the synchronous wave message as well as the code for the neural axon messages and how they might relate. As was shown in the article of immune and nervous system, wireless and intricate local chemical communication is occurring constantly all along the axon and this ephaptic communication along the axon is the basis of the wireless immune system’s intricate and constant communication from neurons to perform immune functions.

Dreams of the Neural Code

The news is filled with dreams of finding the neural code and mapping the brain if enough money is spent on research. This may be correct, but there are many reasons to think that it is too early in brain research to know how to do this.

The popular science press is filled with correlations of brain regions with cognitive tasks, thoughts and feelings. Most scientists would also think that these are accurate, since they are reported by neuroscientists. Along with this are the constant reports of technology knowing how to build a brain with computer technology.

Recently, a leading neuroscientist noted in Nature that:

“The general public might think that this goal has already been achieved; when they read that a behavior is associated with some part of the brain, they take that statement as an explanation. But most neuroscientists would agree that, with a few notable exceptions, the relationship between neural circuits and behavior has yet to be established.“

More recently, in Nature Reviews Neuroscience, it was noted that most neuroscientific studies have too small a number of subjects to be absolutely certain of the results. It notes “that the average statistical power of studies in the neurosciences is very low. The consequences of this include overestimates of effect size and low reproducibility of results.”

A similar recent sentiment was given in Scientific American, “An Epidemic of False Claims” where it is stated:

“False positives and exaggerated results in peer-reviewed scientific studies have reached epidemic proportions in recent years. The problem is rampant in economics, the social sciences and even the natural sciences, but it is particularly egregious in biomedicine.”

Misinterpretation of fMRI is a Big Problem

As mentioned before, there is little proof that blood flow equals neuronal activity (especially since it is determined by astrocytes) nor that it predicts mental activity. But, a recent study questions the interpretations of many fMRI studies for other reasons.

Bold MRI

A recent study has shown that the signals from blood oxygen level–dependent contrast (BOLD), which is the most commonly used fMRI technique measuring blood flow, are very complex and involves activation of at least neurons and glial cells.

There is increasing evidence that studies from MRI and other imaging devices have over interpreted results.  One study, looked at more than a thousand different people (most studies have small numbers), and a second study studied actions repeated more than five hundred times (most studies only repeat a few times).

These two high quality studies showed that interpretations are much more complex. Most events that previously would have been considered localized in one area actually showed many different varied effects throughout the brain. These many small events scattered in wide regions throughout the brain could not be singled out for importance. It was impossible to actually distinguish active and inactive regions. These studies show that previous research only shows the strongest regions, not necessarily the relevant regions.

Another issue raised by these studies is that previous studies noted a particular moment related to the mental event. In fact, there were many more complex sequenced events occurring before and after every event studied. Most current studies did not pick up the complexity.

Conclusions from these studies are that our current tools are just not adequate (which is why the U.S. focus on creating new tools is very relevant). For extreme detail the tissue has to be dead, or in animals. All current imaging tools don’t come close to the millisecond or faster times necessary to observe cellular processes and neuronal activity across the entire brain.

The Complexity of Searching for the Neural Code – Neuroplasticity, Multi Sensory Brain

Previous posts have shown the research that demonstrates that neuroplasticity occurs not in one region but in many far ranging brain regions at once. Studies of connectivity show that the hubs once thought to be isolated modules that communicate are, in fact, each very connected in multiple ways.  Regions that used to be considered for individual senses, such as vision or hearing, are now known to be at least 50% associational, that is connected with many other senses and regions. The fact is that most neurons in the brain connect with multiple senses and modalities.

The more the organization of the brain is studied the less there is known about any central place that creates the sense of “I”, or the unified subjective experience

But, even more confusing is the completely unknown relation between wired brain circuits and wireless circuits of several types. One wireless communication occurs between the immune cells and the neuron. Another occurs with synchronous coupled brain waves sending information from one region of the brain to another.

Another problem not yet mentioned is that it is possible that individual neurons represent, or are used, as a concept such as a “grandmother”. If there are concept, or grandmother cells, than it makes finding a neural code between neurons even more complicated. It is possible that neurons could  be used for multiple different concepts at different times, but this is highly theoretical.

This Week in Science Magazine

Just this week in Science, the Director of National Institute of Health and National Institute of Mental health and the Director of the National Institute for Neurological Disorders and Stroke summarize the monumental difficulties:

“Brain function is a dynamic process with changes on a millisecond scale, but some of our most powerful current mapping techniques are static. Neural circuits involve at least 106 cells in a complex, recursive network, but neurophysiology has been based classically on single-cell recordings or, more recently, on small ensembles of cells. Human neuroimaging captures the whole brain in action, but each 1-mm3 voxel includes at least 80,000 neurons and 4.5 million synapses. An fMRI scan, with 680,000 voxels, is capturing local changes in blood flow and oxygen consumption—but these changes are low-resolution and slow surrogates for neuronal activity.”

Complexity in Searching for the Neural Code

tumblr_lzrva5RJQ91qfjvexo1_1280But, the biggest problem with the computer model of the brain is that the information transformation is occurring on many different levels at the same time – at least 12 orders of magnitude. Mind and brain events occur in:

  • Interactions with other people and culture
  • Entire brain
  • Regional hub of neurons or a small subset of neurons at different times
  • Individual neuron with surrounding glial cells and extracellular matrix
  • Cellular organelles inside the large complex neuron 
  • Inside an organelle
  • Along the axon and dendrite,
  • Inside the cytoskeletal molecules that form the basis of neuroplasticity structural changes and are the language of shape in the neuron
  • Signaling cascades from membrane to nucleus and all through the cell
  • Nuclear pores
  • Chromatin structures surrounding the DNA with epigenetic markings
  • DNA
  • Jumping Genes
  • Markings on the DNA
  • Non coding small and large RNA
  • Alternative RNA splicing
  • Folding and shape of the protein
  • Quantum biological effects

All of these events are occurring simultaneously and in fact represent the same event at different levels of scale over a range of orders of magnitude. (In comparison 12 orders of magnitude from our human body includes the solar system.) And if the paradigm discussed in these posts is correct that mind exists throughout nature, the number of orders of magnitude is far greater.Markram_brain_illos2

As a scientific theory, considering the mind as created by brain molecules, or neuronal switches and circuits, does little to explain these multiple layers where mind exists simultaneously. A much better scientific theory, which has a much greater chance of understanding these vastly different scientific levels affecting the mind and being affected by it at once, is that mind exists throughout nature and interacts at all of these levels.