Is there a real difference between the brain hemispheres?
Liam Keating discusses this important subject in "Associative and oppositional thinking", the paper he wrote for Dialogues in Philosophy, Mental and Neuro Sciences which will be published in a shorter version in december 2017.
I choose to share this full text paper because I think it can be an interesting food for thought about binding problem, that is the problem of how it is possibile to "bind" or to integrate the groups of neuronal activity.
I take this opportunity to thank Liam Keating for hiss willingness.
Now, to understand how brain hemispheres behave, the only thing left is to read!
ASSOCIATIVE AND OPPOSITIONAL THINKING:
THE DIFFERENCE BETWEEN THE BRAIN HEMISPHERES
Author: Liam Keating
Abstract
Concepts need to be separated from each other in the brain in order for an animal to act on one object in isolation. A possible solution is to inhibit common feature neurons that are shared by concepts. One problem with doing so is that each feature of every concept is a commonality with some other concept. This can be circumvented if common features are only temporarily inhibited while two concepts are both active at the same time. The difficulty can also be resolved if the brain is divided into associative and differential or oppositional areas which operate concurrently. Associative thinking, proposed to be the sole mechanism in the right hemisphere, produces poorly defined but well related concepts. Oppositional thinking, speculated in the left hemisphere, cancels out the common features of concepts to expose the differences, resulting in clear definition of the concepts, allowing more accurate targeting of real objects for action upon them. This paper discusses how the proposed operational differences could cause the behavioural differences that have been found in the two brain hemispheres. The theory also has implications to how concepts can be defined mechanically in the brain.
Keywords
Conceptual thinking, Cognition, Binding problem, Lateralisation of the brain, Object manipulation, Spatial perception, Language, Abstraction.
Introduction
What is commonly known in neuroscience as the ‘binding problem’ could more accurately be called the ‘separation problem’. The question has generally been viewed as a problem of how the brain can divide the world into concepts by binding the correct feature neurons together to form each object concept. However, an associative brain that works by strengthening cell connections should have no difficulty in binding feature neurons to form a concept. The feature cells are potentially any cells that are activated by sensing of a colour, shape, sound, smell or other sensation. When active at the same time, it is thought that feature neurons reinfo rce synapses from other active neurons so that they become reminders for each other in future – a network memory or concept.
The problem is that all the feature cells that are active at a given time would reinfo rce together, joining everything that is currently being sensed and memorised to each other. Also, because an object will have some features in common with other objects, these common feature neurons will join concepts of objects that were seen at different times. Therefore in a purely associative brain all the sensory neurons would form a single concept of all that has ever been experienced and all the motor neurons would form one habit or very confused skill.
To some extent objects can be segregated by their independent movement and separability from other objects. Vision is differentiated by tactile knowledge and vice versa. Also motivations, desires and fears that target specific features, can provide top-down definition of the world by limiting an object to its features of interest. Furthermore, the overlay of features in a single location, for instance a shape and colour in the same retinal area, can correctly be bound together in the retinotopic or somatotopic regions of the cortex and these regions might coordinate binding of the same features in non-locational, non-'topic cortex areas.
However there remains a difficulty of how two object concepts that contain a common feature neuron can be separated. The common cell will bridge the boundary and a desire or fear complex will not be able to distinguish which feature neurons form the concept that contains the features of interest and which neurons are only an object associated with a wanted container object. In other words, without some way to segregate the concepts, the motive to eat (or avoid) will drift from the object, that contains the desired sensory features, to the associated object, as long as they both contain some of the same features. The animal will eat the wavy looking meat and the wavy looking soil, the red meat and the red flowers it is near to. It would not be able to segregate the desired from the undesirable in its memory, therefore nor in its action targetting.
This paper presents a theory to explain the different behaviours of the brain hemispheres. The mechanism that is speculated to operate in the left hemisphere can also explain how the world is sliced up to form isolatable object concepts. In brief the idea is that in the left hemisphere any common feature neurons shared by two active concepts are inhibited so that the shared features cannot bind the two concepts.
For completeness I want to add that there is a separate question that can rightly be called a binding problem. That is, how can you know that a blue patch or circle seen by cells on one part of the retina is the same as a blue patch or circle seen by cells in a different retinal region? Or how can you know that a pin-prick or warm touch to the shoulder is the same quality as a pin-prick or warm touch to the foot? The simplest solution would be for each of the retinotopic and somatotopic neurons that register an identical feature to be connected to a hub neuron. The hub neurons, with connections between each other, could form non-locational concepts. This is explained slightly more elsewhere.
Brain hemisphere behaviours
The right hemisphere (and left hand) holds an object steady, the left hemisphere (and right hand) operates the tool. The right hemisphere seems to memorise relationships between objects while the left memorises components within objects. This is suggested by how the right hemisphere of split brain patients can draw a simplified house with no detail whereas the left can draw the parts in detail - roof, chimney, windows etc - but in the wrong places (Gazzaniga, 1967; Lamb and Robertson, 1991; Bogen 1969). The right hemisphere is more proficient at map reading and other spatial tasks. The left deals with most speech and language but struggles to recognise or speak connecting words such as 'be' 'and' 'or' 'of' (Lehrer, 2012). The right can understand language, including previously unheard metaphors, but seems unable to speak most words (Altmann et al, 2012). The left hemisphere has been generally found to be the impulsive, incautious side, whereas the right tends to be involved with retreat or wait behaviour (Carmel
The right hemisphere might not be able to recognise boundaries. Jill Bolte Taylor, a neuroscientist, gave a TED Talk about her experience during a stroke that affected the left side of her brain. For some phases while she sought help she said that she could not dial a phone number and presumably only her right hemisphere was operating then. At these times she looked at her arm and at the wall behind and said that she could not conceptualise where her arm began and the wall ended (Taylor , 2008).
It is difficult to discern a common thread in these hemispheric differences. Overall it seems that the left hemisphere isolates objects to act on them whereas the right hemisphere connects things to see wider relationships, allowing it to recognise wider consequences and allowing it to hold an object steady in relation to other things.
Proposed mechanisms
How can the left brain isolate and distinguish two object concepts? When two different but similar concepts are live in the brain at the same time, any feature cell that they share will be over-active compared with the other feature cells. This is because the shared feature is powered by the sources of both of the live concepts that it is a part of. The concepts may be live because one concept is being searched for and the other is seen in reality, or because they are both being searched for and therefore presumably energised by subcortical motivation/search structures.
In any case, if the second live concept is different from the primary concept that is wanted and the two concepts share one feature, then this over-active feature cell would reinforce strongly to the other feature cells of both concepts. The common feature cell will act as a link that activates both concept networks in future, whenever one concept is active.
As a simplified example, imagine a young scavenger animal that has experienced meat and found it satisfying but so far only differentiates it by its redness, as it has seen meat of many shapes and textures but always red. Then it eats some liver from a carcass. Let's assume that the new taste wakens an avoidance instinct (see note 1 at the end) that deems it a poison. The animal cannot differentiate liver by its colour which it shares with meat but the different visual texture can distinguish the two - smooth appearance of liver versus fibrous or wavy meat appearance. The animal licks and checks the visual appearance of one part and concentrates to reiterate the signal and reinforce the association between liver taste and smooth visual texture. Then it does the same to further memorise meat and its now more limited visual texture range.
However, the association between red and meat and red and liver taste will also be strengthened with each iteration. To discern and to memorise liver and meat as two concepts when they are both in view, to treat them differently, the brain must inhibit activity of the link cell, the 'red' cell (see note 2).
It cannot do this simply by inhibiting whichever is the most active cell as the shared feature cell will not necessarily be the most activated compared with other prominent feature cells. However, the shared cell will activate more strongly than usual when two concepts that include it are both active and this would allow the common cell to be identified.
The searching instinct or subcortical complex cannot trial to find the unusually active feature cell by turning off one concept at a time, since a comparison cannot take place while the wanted concept is inactive.
Therefore, does the brain distinguish the anomalously energetic feature cell by measurement of feedback? Does it measure its supply of power to the feature cells of the wanted concept and then measure and compare the returning signal from each of those cells? A type of radar in the brain? It could then distinguish if one feature feeds back more than it was supplied and must therefore be shared by another live concept.
A concept set of features is lit up when it is searched for. A different object may be found in reality that has one feature in common with the wanted concept. The searching complex supplies energy and measures feedback from the wanted feature cells and one cell returns more power than all the others. The left hemisphere inhibits this cell. The inhibition of disproportionately elevated feedback in the left brain will split two active concepts where they share a common feature. This could explain many of the differences in behaviour of the brain hemispheres, as described further on.
When a fully matching object is found in reality, the wanted concept cells will all feed back more energy than they were supplied. No cells will be disproportionately active and the uniformly increased signal from all wanted feature cells registers the discovery of the wanted thing, by switching the brain to approach behaviour.
Sometimes an animal will need the left brain to expand its associations to an object. For example if food is always found next to a tree trunk, then the food-trunk association would help in future searches. Therefore perhaps the inhibitory mechanism only operates when the subcortex is actually enervating a concept, searching and testing it against things in reality. When not actively searching, memories can form in the normal associative way, in proportion to real encountering.
***
Let's assume the same type of concept-commonality highlight mechanism serves the right hemisphere, except that when a cell feeds back too brightly it is energised more instead of being inhibited, to strongly reinforce its connections to other active cells. The shared feature in the right hemisphere therefore becomes the main feature of the relationship between the two partly similar objects.
The difference then is that the right hemisphere will tend to shift focus towards wider and larger relationships, category memberships with few commonalities, whereas the left will tend to focus on components. This happens in the left brain because when doing almost anything with an object, you have to focus on it one part at a time. Since for the parts, the whole is a commonality, the whole disappears to the left brain. This inhibition will apply to visual cortex cells that are activated by the sight of lines or edges at the boundaries of two things and also to the complex shape cells that register the shape of the whole.
This can explain why the left brain can copy parts of a drawing but fail to draw how they relate as a whole. It can also explain why the left is poor at map reading and other types of spatial cognition, as it inhibits the link between adjacent components whenever concentrating on one. The right brain simplifies by concentrating on the common junctions between parts of a drawing, therefore conversely mainly ignoring the details.
***
The face recognition centre is typically on the right side in the human brain, in right-handed individuals. If the theory postulated here is correct then the right hemisphere should be no better at discernment and perhaps worse than the left. The left brain will distinguish faces by their internal relationships and compare one face to other faces by cancelling out similarities, to narrow down the distinguishing features of a specific face.
The right brain will distinguish weakly by internal facial features as it has no mechanism to differentiate while two faces are live in the mind. However, the features of a single face on its own in reality will reinforce to many other memories that involved that person. The whole persona will be recognised so the visual feature cells will also be somewhat distinguished by reinforcing strongly to the contextual memories. Individuality, a strong idea, in the right hemisphere is not difference but breadth and strength of links to other memories. Facial individuality will form there because one face is associated to one set of memories, a different face to a different set.
However, commonalities between faces will cause the right brain to expect common behaviour. This will make the right-hemisphere better at reacting to expressions than faces. It can still work well enough for face recognition because the more memories a person is associated with, the more highlighted and therefore differentiated are their facial features.
The reason the right hemisphere has difficulty speaking may be because it connects words to their opposites. In the right brain everything will mean everything else to some extent. Opposites such as the concepts of 'man’ and ‘woman’ have more similarities than differences. Each word is slightly more connected to its respective real gender group but both words will also be cross-connected in the right hemisphere by their commonalities - both words mean human, adult, hairy in some places etc. The feature networks of both words - sound or visual letter features - reinforce to both gender concepts so the word definition is poor. Such definition becomes strengthened, the dent of a word and meaning association is deepened, by regularity of the word being used while one gender is present in mind. With strongly associated words the right brain could still not exclude other words but could guess with some confidence and could possibly learn to speak by this means.
When the right hemisphere is asked to specify a word for something, many related words will come to mind including contextual opposites and category options which would include some inappropriate words from higher categories. The right brain will find it difficult to choose. However, when the right brain is spoken to, the words will connect to their respective concepts more than to others, so meanings can be grasped. Also the words of a sentence that have been heard will affect connectivity of each other's concepts, so the sentence altogether supports a possible meaning with sufficient confidence for the right brain to believe that it understands what was said to it.
***
To sum up, lone-highlight inhibition in the left hemisphere will result in division by commonality and therefore to see the thing that is focused on as an isolated object. In the right hemisphere reinforcement of an anomalously active cell will allow a concept to be related to others that share that feature.
This happens when the brain searches for an object in reality by energising the concept it wants. If a common quality is seen in a different real object, the left hemisphere will note the difference, the right hemisphere will be tempted by the similarity.
Also it happens when searching for a commonality and not a specific object, for example when looking for anything sweet or anything that can act as a screwdriver. Then the right brain comes into its own.
Further considerations and questions
Concurrently experienced objects versus timeless commonalities
If two objects are present in the same place and time, all feature cells of both objects can associate to all others. This is labelled here as a concurrent association (CA).
If two objects are not experienced together at the same time, only the features that they share will link the two network concepts. This type of concept relationship is called here a timeless associate (TLA).
When the brain searches or thinks, I assume that it intensifies the signal through whatever concepts it wants to find or think about. The signal will pass from the activated concept to the feature cells of its CA's ‘all to all’. Each of the wanted concept’s feature cells will also be associated to other concepts with a category link, one common feature, the TLA's.
Presumably the brain or subcortex only knows it has found an answer by feedback from the quest concept itself. When thinking, there is no second concept lit by reality, so the only feedback will come from loops where the signal has passed through to an associated concept network and back again to the activated feature cells that supplied the loop.
If a CA loops back, the activity of all the features of the wanted concept will be increased equally so the left brain will not inhibit the CA. However if a TLA receives activation through a single shared feature, and the signal loops and returns rather than dissipating, then that feature will stand out. The left hemisphere, while in search/anticipation/thought mode, will detect the anomalous high feedback and inhibit the overactive feature cell that is feeding the TLA.
The left will remember things that were actually seen together. It would have no access to memories of all the objects that share one or a few features with a wanted concept. Category knowledge will only partially and accidentally form in the left brain because of being told, within a few seconds, that pineapples and tomatoes are both fruit, or that sunsets are similar to tomatoes in colour, or if at some time tomatoes were actually looked at (while in idle mode) during a sunset.
If asked "What is the capital of France?" the left hemisphere enervates all features of both concepts. The signal passes through many concepts that were heard or otherwise experienced together with 'capital' and 'France' in the past. The 'Paris' concept feeds back without anomaly as a CA to both question concepts, returning power to all the feature cells of both, because it has been experienced with both many times. This evenly increased feedback registers as a match and causes the left brain to trigger speech.
How can looping cause the signal to increase and produce feedback, rather than to dissipate? Does there have to be some mechanism, perhaps a specific neurotransmitter released while in anticipation/thinking mode, that encourages cells to reach action potentials when receiving weaker inputs than usual?
Anomaly ratio
The signal ratio of [anomalies]-versus-[all feature cells] that triggers the inhibition mechanism must be small so that if enough anomalies are registering, the object can be recognised as a match that is partly hidden or viewed from a new angle. Also, a concept is not a set of pinpoints in feature ranges but is whole sections of each range. For example meat can be various textures and colour shades. A real sample of meat will only activate short sections of the remembered concept ranges, so the trigger anomaly ratio has to be very small.
Either that, or the whole concept ranges are at first searched for, then when a piece of meat is found, the smaller, visually active feature range sections are focused on, then the anomaly mechanism operates.
Abstractions in left and right
In the right hemisphere, all is abstraction, every object is made of features that relate it to other objects that have one of those features. In other words, all is abstract categories in which tomato and sunset can be classified together just for both being red, even if never seen together.
In the left hemisphere, operations can be abstracted. For example, if the left brain is told that "A is taller than B and B is taller than C" and then asked "Is C taller than A?" it will inhibit 'tallness'. This is because each object has tallness as a common feature. Except for the nouns, the words that remain are 'is ____er than'. Individually these two words and suffix are likely to be difficult for the left brain to understand, since they have no opposites against which to form boundaries, no negatives to define them except by adding the word ‘not’. Altogether they represent a common correct grammar and they mean 'more', signified by the '____er'. The 'is' and 'than' signify the direction.
This is a left brain type of abstraction. A descriptive operation without a description, it is replaced by a silent generalisation of all adjectives that act with grammar the same way.
In a real comparison of tallness, when a subject is asked to say which of two lines on a page is taller, 'tallness' cannot be inhibited. The other features of the two lines will be inhibited in the left hemisphere. The concepts of 'tallness' and 'most' have to be assisted (by an assumed subcortex module) since they are the search concepts, so the two lines become two simple features, vertical lines in two spatial locations.
How tallness of the lines is actually compared is unknown, perhaps achieved by linking to old memories of making such comparisons and the valuing of the 'most' that they contain, by the memory of some type of reward. Or perhaps by taking advantage of the graduated nature of each feature cell range, by measuring the signal strength from one end of a feature range and comparing the signal that is produced when looking at each of the two lines. In this instance the compared signal would be from the 'tall' end of a range of visual cells that are activated by different vertical lengths.
Motives determine the focal point
Desire or fear determines what is an object - that contains an instinctive, pre-wired cue-set of features - and what is an associate or background. Clearly, or unclearly, an instinctive cue-set is contained in the object but also within the environment or background that contains the object. In the left hemisphere the desire or fear draws limits, for example where a desired taste is found to geographically overlap with a visual feature, this appearance is meat taste, all else is cut off from the concept due to either lacking commonalities or having too few. In the left brain desires and fears cut reality into pieces. In the right brain a desire or fear is more like a hammer, it creates a definite centre - a pre-wired feature cell network - but with an indefinite surround of associations, that are weaker and weaker at greater distance from the central instinctive concept.
Possible mechanisms
No doubt there are many possible architectures for the match/commonality/non-match mechanism postulated here and it is not the aim of this paper to specify the probable architecture. The mechanism would need to include a method to record the signal strength issued to a feature cell (or feature cell range) and then to somehow compare the return signal against what was supplied. Perhaps the recording is achieved simply by activating a chain of cells at one end such that the stronger the signal, the further it progresses along the chain. The return signal then has to be tested on the same chain as the supply signal for a comparison. The results from different feature cells would have to be collated, with different decision routes depending on the proportion of feature cells that feed back excessively.
Concluding comments
The mechanism proposed here can explain many of the reported behaviours of the left and right hemispheres. Questioning of subjects with right hemisphere damage shows that the left brain performs poorly in understanding incongruent adjectives in sentence pairs, such as 'John is taller than Bill. Who is shorter?' (Caramazza et al, 1976). This is explainable because the main definer and therefore regularly inhibited partner of 'tall' in the left brain is 'short'. As the silent partner, unreinforced at the commonality, the opposite word cannot be accessed while thinking except by a circuitous route.
The theory can account for why the left tends towards the obvious, the things previously encountered together in reality or speech at the same time, while the right tends towards rare commonality, higher category, timeless associations (Beeman et al, 1994; Burgess and Simpson, 1988).
The right is subtle and the left is precise. The right is dense but vague, the left is definite although something is always neglected. In the right hemisphere the visual edge/line cells at the boundary between Jill Bolte Taylor's arm and the wall behind joined them as one concept, while Jill was suffering a stroke. In the active left hemisphere the edge/line cells are the inhibited commonality, the division of adjacent things.
Both hemispheres have full concepts formed by experience of many examples of dogs or towns or of a particular face. In both hemispheres associations form between concepts that regularly occur together. Both hemispheres can distil a person to their component organs and atoms. There would have to be a motive to cause such distilling. The left hemisphere would see the organ in the person. The right hemisphere would see the person as one of many with such an organ.
To communicate, the right hemisphere points, and it may be trying to indicate the whole landscape, the forest in the middle, or just one tree. The left hemisphere has to gesticulate to exclude what it does not mean, and so it invents oppositional language. Most concepts and words are defined both associatively and oppositionally, for example 'apple' is associated to many examples of apples you have had and is also defined oppositionally by everything that is not an apple. It is poorly defined against jetskis and carpets, more precisely defined in opposition to peaches and pears.
Existence and consciousness are examples of concepts that are mainly defined associatively. Their closest opposites are non-existence and unconsciousness, which are too absolutely different from the positive concepts to add much definition. In any case these opposite concepts cannot be experienced. The concepts 'existence' and 'consciousness' associate to all experienced or thought of things, so even associatively the two concepts can have little definition.
One prediction of the theory set out here is that the left hemisphere should have difficulty with the concepts of 'consciousness' and 'existence', in the same way that it apparently has difficulties with shorter words that cannot be defined oppositionally. The prediction is weakened since the left brain might have an oppositional understanding of consciousness, by the third person view of other people sleeping or awake. It is hard to see how the theory might become firmly established other than if a suitable mechanism is found. However the brain must do something like this.
Note 1
An instinct is envisaged as a sensory trigger that when activated connects on to an action or search procedure. For hunger the trigger is sensors in stomach and blood. For most instincts the trigger is a pre-wired fear image such as a pigeon's fear of a hawk, or a desire image or tactile concept, for instance to direct newborn suckling or adult mating.
Note 2
Colour is coded initially in the visual cortex as patterns of activation across small lumps of many cells known as ‘colour blobs’, that are each activated by small patches of retinal cone cells. However, it would be more useful for colour to be represented by single cells on a spectrum, each activated most by a specific pattern of activity in a colour blob. This may be the case further forward into the cortex. Single cell registers would allow less complicated memory formation with other feature cells than if using colour blobs.
References
Altmann, U., Bohn, I.C., & Jacobs, A.M., 2012. Looking at the brains behind figurative language - a quantitative meta-analysis of neuroimaging studies on metaphor, idiom, and irony processing. Neuropsychologia. 50(11):2669-83.
Beeman, M., Diamond, S., Friedman, R.B., Grafman, J., Lindsay, M.B. & Perez, E., 1994. Summation priming and coarse semantic coding in the right hemisphere. Journal of Cognitive Neuroscience, 6(1), 26-45.
Bogen, J.E., 1969. The other side of the brain. I. Dysgraphia and dyscopia following cerebral commissurotomy. Bull Los Angeles Neurol Soc. 34(2):73-105. Weblink showing drawings: https://www.nobelprize.org/nobel_prizes/medicine/laureates/1981/press.html
Burgess, C. & Simpson, G.B., 1988. Cerebral hemispheric mechanisms in the retrieval of ambiguous word meanings. Brain Lang. 33(1):86-103.
Caramazza, A., DeLuca, D., Gordon, J. & Zurif, E.B., 1976. Right-hemispheric damage and verbal problem solving behavior. Brain Lang. 3(1):41-6.
Carmel, D. & Grimshaw, G.M., 2014. An asymmetric inhibition model of hemispheric differences in emotional processing. Front. Psychol. 5:489.
Federmeier, K.D. & Kutas, M. 2011. Thirty years and counting: Finding meaning in the N400 component of the event related brain potential (ERP). Annu Rev Psychol 62: 621-647.
Gazzaniga, M.S., Hillyard, S.A. & Kutas, M., 1988. Processing of semantic anomaly by right and left hemispheres of commissurotomy patients. Evidence from event-related brain potentials. Brain 111 (3): 553-576.
Lamb, M.R. & Robertson, L.C., 1991. Neuropsychological contributions to theories of part/whole organization. Cogn Psychol. 23(2):299-330.
Lehrer, J., 2012. The neuroscience of Bob Dylan's genius. Interview with Mark Beeman. The Guardian, 6th April
Taylor. J.B., 2008. My stroke of insight. TED. Available at: https://www.ted.com/talks/jill_bolte_taylor_s_powerful_stroke_of_insight
