Semblance Hypothesis

After more than a decade of examination by adhering to best available scientific methods1-6, mounting evidence forces me to regard semblance hypothesis as a theory. Despite several open invitations to disprove the hypothesis through both this website and a large number of scientific presentations and peer-reviewed publications, no objections were received. This is a theory of nervous system functions that provides testable predictions (pdf with methods to test them). I sincerely hope that scientific community will use the time-tested method of "testing the predictions of a theory"7 with an aim to disprove it. Please explain the importance of this to your community leaders and policy makers. I thank all those who have supported me during several difficult times of its development - Kunjumon Vadakkan, dated 21st March, 2020


                  1. Strobel N. Method for finding scientific truth. Website

                  2. Strobel N. What is a scientific theory? Website

                  3. Goodstein D (2007) A testable prediction. Nature Phys. 3:827 Article

                  4. Lee AS, Briggs RO, Dennis AR (2014) Crafting theory to satisfy the requirements of explanation. Article

                  5. Lee AS, Hovorka DS. (2015) Crafting theory to satisfy the requirements of interpretation. Article

                  6. Dutailly JC (2017) Chapter 1. What is science? Theoretical Physics. p1-24.

                  7. Bialek W (2018) Perspectives on theory at the interphase of physics and biology. Rep. Prog Phys. 81(1):0126001 Article


 by Kunjumon Vadakkan 


Objective: To understand how first-person inner sensations (in the mind) of higher brain functions (such as memory & perception) occur both independently & along with third-person observed motor activities such as speech and behavior.



Dedication: To all those who suffer from diseases of the brain (& therefore mind), especially who are abandoned by their families.


Reaching the solution for the brain

How to understand something that cannot be accessed by our sensory systems?   A method used in physics

How can we use a non-real intermediary to solve the nervous system? An example of solving a problem that geometry fails to solve

A deep principle useful for solving the nervous system - demonstrated by an example

A medication for unrelated neurological and psychiatric disorders - What does it inform us?

Insulating extracellular matrix - How thick it is? Can information get etched on it?

Has learning-mechanism got features of an evolved mechanism?

There is no need for a separate mechanism for working memory

How is learning related to LTP induction? An explanation

Extreme degeneracy of input signals in firing a neuron

Does the brain do retrograde extrapolation?

Importance of triangulation in verifying a mechanism 

Testable predictions made by semblance hypothesis

Perception from a first-person frame of reference

Without sleep, there is no system! An explanation

Internal sensation - A comparison with electromagnetism


      If you are new to this website, then please don’t read content in this box first. Please read about the hypothesis and come back here.

The basic concept of “attention heads” in Transformers matches with partial features of “islets of inter-LINKed spine heads,” a model of nervous system functions


The semblance hypothesis has put forward a model of operations of the nervous system. Here, spatial relations between specific signals arriving from input neurons are registered in “islets of inter-LINKed spine heads (IILSHs)”, whose spines belong to different output neurons. In response to a prompt (cue stimulus), operation of IILSHs generates both first-person inner sensations and motor outputs such as speech and behavior. The field of artificial intelligence uses artificial neural network where neurons are connected with each other in different configurations. The configuration in large language models (LLMs) has succeeded in showing features of good generalization - ability of a trained system (nervous or artificial) to perform well on new input data unseen during training. This prompted to examine the basic concept of operation in the LLMs. This shows that "attention heads" within the hidden layer located in between input and output neuronal layers of Transformer in LLMs is equivalent to a linear algebraic treatment of one segment of operation of "islets of inter-LINKed spines" that generates motor outputs. Article


If a primary school child ask me, I will tell this work as a STORY

     This is a video uploaded on 8th July, 2023. If you are new to this website, then please read this front (home) page and FAQ section of this site before watching.  

An explanation for associative learning in terms of semblance hypothesis


Video  (Note: Qualia is a virtual first-person property. Qualia of inner sensation can be estimated by a retrograde extrapolation from postsynaptic terminal D towards the sensory receptors. Even though it is shown as part of the video for demonstration purpose, there is no back propagation of potentials in that direction. Generation of inner sensation is a system property of systems that has oscillating extracellular potentials whose vector components are contributed by synaptic transmission across synapse A-B & propagation of potentials along the inter-postsynaptic functional LINK B-D in a perpendicular direction to the synaptic transmission).

This video provides an explanation for changes that occur when sound & foot shock are associatively learned. It then explains how sound alone (after learning) triggers inner sensation of memory of foot shock & cause foot withdrawal. Neutral sound is the conditioned stimulus (CS). Aversive foot shock is the unconditioned stimulus (US). Both reaches neurons of Lateral Amygdala (LA). Certain learning-change is expected to occur at locations of their convergence in LA. After associative learning between sound & foot shock, sound alone causes an inner sensation of foot shock & triggers foot withdrawal. A testable explanation for a link between pathways of sound & foot shock where they converge in LA is expected. At the location of convergence, sound should generate inner sensation of foot shock (in the absence of foot shock). This is like a sound-induced hallucination of foot shock. Marvin Minsky gives us some guidance -

Questions are a) "Is there a suitable location where hallucinations can be triggered by sound during memory retrieval?" b) “What are the pre-conditions that set a stage to trigger hallucination of foot shock when sound arrives?” This can be explained using two examples. a) First is a method by which a veterinarian gives an intravenous injection to a horse. Initial patting on the side of neck (over the visible internal jugular vein) followed by tapping with increasing intensity enables the veterinarian to rapidly place a needle in between the fingers of the tapping hand to get inserted to that vein without notice. b) Second is about pickpocketing. It is easy to get pickpocketed while climbing stairs (e.g., in a railway station) since the gluteal region is frequently moving while taking steps. From both, it is possible to insert/remove something to/from a system that is undergoing a continuous sequence of events without the system’s notice. In both cases, newly inserted stimulus is perceived as one of the frequently occurring events. i. e., the newly inserted stimulus leads to a hallucination (at the location of insertion) that it is receiving a usual stimulus that has been occurring prior to it.

A postsynaptic terminal (spine) is continuously being depolarized by quantally released neurotransmitter molecules (NTMs) from its presynaptic terminal. When an action potential arrives, the latter releases a volley of NTMs to trigger a large postsynaptic potential. For a postsynaptic terminal (spine), it receives inputs from its presynaptic terminal all the time & sometimes from the environment via its presynaptic terminal (until learning occurs). This is the dominant state of the system. In the above contexts, if learning cause formation of an electrical link between spines (that belong to different neurons, since spines on a dendrite are separated (mean value of inter-spine distance) from one another by a distance more than the mean spine diameter) to which associating stimuli arrive, then we can examine whether it will facilitate a mechanism for memory retrieval. Later, arrival of one of the associatively learned stimuli through the link that reaches the lateral aspect of the spine (and depolarizes the entire spine) to which the second stimuli had arrived during learning will be perceived as if the second stimulus has arrived from its presynaptic terminal. This forms the basis for a hallucination (as Marvin Minsky envisioned). In other words, depolarization of a spine from its lateral aspect tricks the latter to hallucinate that it receives an input from its presynaptic terminal (from the environment).


Since mean inter-spine distance is more than mean spine diameter, the above link is expected to occur between spines that belong to different LA neurons. The potentials elicited by sound propagates through the link to reach the postsynaptic terminal through which shock arrived in the past to trigger hallucination (inner sensation) of shock. Potentials then propagate further to trigger foot withdrawal. There should be a mechanism to integrate units of inner sensations originating at many nodes of convergence to cause inner sensation of foot shock. The brain functions only in a narrow range of oscillating extracellular potentials. A cue stimulus reaching the link leads to propagation of depolarization in perpendicular directions - through synapse A-B (see video) and link B-D. These can provide vector components to oscillating extracellular potentials. Thus, we have a mechanistic explanation for a testable change that occurs during associative learning & a mechanism for cue-induced cue-specific hallucination that generates inner sensation of foot shock & foot withdrawal.


     If you are new to this website, then please don’t read content in this box first. Please read about the hypothesis and come back here.

A new explanation for flash-lag effect using a testable mechanism for first-person inner sensation of perception

When a flash is briefly presented in a specific location adjacent to the path of a uniformly moving object, the former is perceived to lag the latter. Different experimental conditions carried out to qualify this flash-lag effect (FLE) have led to several seemingly unrelated first-person reports from which several constraints are available to synthesize a testable mechanism for FLE. Using a derived mechanism for the generation of first-person internal sensation of perception, present work provides new interconnected explanations for disparate findings associated with FLE. By verifying testable predictions arising from the hypothesized mechanism of first-person inner sensations, it may become possible to understand timescale-matched neurobiological changes responsible for perception. Article

If you are new to this website, please don’t read content in this box first. Please read about the hypothesis and come back here.

Golgi staining of neurons: Oxidation state-dependent spread of chemical reaction identifies a testable property of the connectome

Abstract: Camillo Golgi observed the reticular nature of the nervous system by his staining method. Ramón y Cajal modified this protocol and obtained staining restricted to individual neurons, which was in agreement with the cell theory. Close examination shows that Golgi used an oxidizing agent to pre-treat the brain tissue before the staining reaction and Cajal used an additional oxidizing agent for the same step. It shows that oxidation state of the tissue has a crucial role in determining the spread of Golgi chemical reaction between neurons. Since a) dye injected into a neuron spread only within that neuron's cytoplasm, and b) it is possible to grow individual neurons in primary culture, oxidation state-dependent spread of chemical reaction should take place through a gate while keeping the cell membrane intact. These constraining features guide towards the solution where formation of such a gate between spines of different neurons is a suitable mechanism that can take place during learning. As long as this gate exists, it should allow propagation of depolarization from a cue stimulus across the gate and provides options to a) generate internal sensation of associatively-learned item, and/or b) provide potentials to cross the threshold to fire neurons to generate motor action reminiscent of the arrival of associatively-learned second item. These constraints lead to the derivation of a crucial connection that can define the connectome. This oxidation state-dependent reversible connection of the connectome matches with the operational functions of inter-neuronal inter-spine functional LINKs proposed by semblance hypothesis. All the above findings are testable. Article


In simple words, what is semblance hypothesis? Systems in the body are being studied by observing them from outside (e.g. pumping of the heart, filtering by the kidneys, structure of DNA and synthesis of proteins). Third-person approaches are suitable for their studies. In contrast, functions of the nervous system such as perception and memory are first-person inner sensations (within the "mind") to which only the owner of the nervous system has access. But we have been studying these functions by examining the nervous system from outside using third-person approaches at various levels (biochemical, cellular, systems, electrophysiological, imaging, and behavioral) and we were trying to find correlations between these findings with an aim to understand the system. By these approaches, the first-person internal sensations of different higher brain functions have been remaining unexplored. Reality is that the examiner should become an insider in a subject's nervous system (and become part of the system) to sense the first-person internal sensations! This is practically not possible. This means we are facing a brick wall in our current approaches to understanding its operations. This can be overcome by undertaking a theoretical approach. Since a universal operational mechanism is expected to be present in the nervous systems of different animal species, we should be able to find it without much difficulty if we are on the right path. We had no previous experience of searching for a biological mechanism that gives rise to virtual first-person inner sensations. This should not hinder our efforts in any way. In fact, we must prepare ourselves to look for a unique mechanism that has the ability to evade our attention! By keeping this in mind, a theoretical examination was carried out to arrive at a specific location where such a unique mechanism can be expected to take place. Further examination of this location has enabled identification of a set of unique features necessary for a feasible operational mechanism whereby the system can generate units of inner sensations at specific conditions (that are physiologically present). This mechanism is expected to interconnect all the findings made by third person approaches at different levels. Until now, results using this observation are able to explain and interconnect a large number of findings from different levels. Pathological changes of this cellular mechanism can explain several neurological and psychiatric disorders. Predictions made by this hypothesis are testable.

Is there a different way to view semblance hypothesis? In order to understand the brain, we must understand how it's unique function – generation of first-person internal sensations of perception, memory, and thought processes is taking place. Associative learning between two stimuli is expected to produce certain changes (in a few milliseconds (see FAQ for explanation)) that allow one of the associatively learned stimuli (cue stimulus) to generate the internal sensation of memory of the second stimulus (also, in a few milliseconds). For this to occur, changes during associative learning are expected to take place at the locations of convergence of sensory stimuli within the brain. Here, we need to ask the following questions, “Is there a possible cellular location where the processes of neurons through which associatively learned sensory input signals arrive can converge and make some signature changes during learning?" "If associative learning can produce certain changes at this location (in a few milliseconds), can it be used by one of the stimuli (the cue stimulus) to generate internal sensation of memory of the second stimulus (in a few milliseconds)?” “At what structural location and by what mechanism does the cue stimulus spark internal sensation as a first-person property?” “What is necessary to spark internal sensation?” “What is the basis of sensory features or qualia of internal sensation?” “What holds the system together so that the internal sensations generated from different locations of convergence of sensory stimuli can allow the cue stimulus to generate first-person internal sensation of the second stimulus?” "How can the mechanism that holds the system together relates to the narrow range of frequency of oscillating extracellular potentials (as evidenced by EEG findings) at which both learning, and memory retrieval take place?" "In other words, is there a mechanism that integrates internal sensations induced at different points of convergence to provide memory?" "How does the mechanism of generation of internal sensations relate to behavioral motor activity?" “Can the derived mechanism be extended to explain different brain functions in an inter-connectable manner?” If we look hard enough, we are expected to find a mechanism that can explain all the above features at the location of convergence of sensory input signals. When an inquiry was made to solve this puzzle, it was possible to derive a solution. This testable hypothesis was named "semblance hypothesis".

Is it possible to explain semblance hypothesis by yet another way? Studies of the nervous system have been facing three major challenges. 1) Due to limitations of methods that can be used to understand memories in their true sense as first-person internal sensations, memories have been studied using their surrogate markers such as speech and behavioral motor activity. 2) Current investigations are primarily based on the following postulate made in 1949 by Professor Donald Hebb. "When an axon of cell A is near enough to excite a cell B and repeatedly or persistently takes part in firing it, some growth process or metabolic change takes place in one or both cells such that A's efficiency as one of the cells firing B is increased" (Hebb D. O. The organization of behavior. New York: Wiley & Sons). Modification of this postulate is generally known as synaptic plasticity thesis. Until now, this thesis was not able to provide a mechanistic explanation of how learning-induced changes are used for the generation of inner sensation of memory. 3) Behavioral markers of memory retrieval are being correlated with the firing of a set of neurons with the hope to understand the mechanism. The fact that only a minor fraction of input signals (nearly 140 input signals that arrive from different locations on the dendritic tree) can fire a cortical neuron having thousands of input terminals (dendritic spines where postsynaptic potentials are generated) (see FAQ for references) shows extreme degeneracy of input signals in firing a neuron (see FAQ for references). Input signals (postsynaptic potentials) attenuate as they propagate towards the neuronal cell body. Since many neurons are being held at sub-threshold activation state at rest, it is possible that even a fraction of one postsynaptic potential can fire a neuron from its resting state. In these contexts, to avoid loss, information storage is most likely taking place using a mechanism occurring at the location of origin of postsynaptic potentials (input signals). This storage mechanism is also expected to provide an explanation of how first-person internal sensation of memory is generated. In the past, we kept certain notions such as 1) operating mechanism is occurring at the synapses, and 2) a neuron is an operational unit of the system. Those notions were necessary to initiate experimentations. Since we have already spent enough time to test those ideas and since we are not reaching towards a foreseeable solution, we should try to find testable solutions to understand how the first-person inner sensations of different higher brain functions are generated. During the last several decades, we have made very large number of observations at different levels of the nervous system. Now, we are in a position to use constraints available from all those observations to try to derive a theoretically fitting testable mechanism that can explain how first-person internal sensations are being generated. If successful, this will provide a solid scientific basis for the mechanism that we are seeking. The mechanism is expected to operate in synchrony with the synaptically-operating nervous system and occur at physiological time-scales of milliseconds. It is also expected to operate in agreement with the observation that there is a huge redundancy of input signals that can fire a neuron. If a hypothesis can explain all the features of the system in principle, then we can start verifying both its structural features and testable predictions. Semblance hypothesis resulted from this approach.

Can we explain semblance hypothesis by using some pictures? The unknown mechanism is shown as a black box in Figure 1 below (Note: After reading this and FAQ pages of this website, one is likely to understand the contents of this black box, which can be subjected to further verification).

                            Black Box of the brain           

Figure 1. How to solve the black box of the nervous system? Let us imagine that associative learning takes place between two sensory stimuli in a learning event. Stimulus 1 and Stimulus 2 activate their corresponding sensory receptors, and the stimulus-induced depolarizations propagate through their synaptically-connected neuronal paths (Neurons along the pathways of Stimulus 1 and Stimulus 2 are marked N1 to N5 and N6 to N9 respectively). We want to know the location & the type of change that occur during learning. From the above figure, it is reasonable to expect certain changes at the location of convergence of Stimulus 1 and Stimulus 2. What is expected is that when Stimulus 1 arrives after learning (as a cue stimulus), it should be able to generate an internal sensation of features of Stimulus 2, which we call as memory of Stimulus 2. From the figure, we can ask "Where does neuron N9, through which Stimulus 2 arrives, meet with the pathway through which Stimulus 1 arrives?" The learning mechanism at the location of convergence should also explain how the cue stimulus (Stimulus 1) can produce behavioral motor actions reminiscent of Stimulus 2. If we can provide explanations for these, then we are moving in the right direction for solving the system. Previous studies examined both 1) changes occurring presumably in all the synapses along the pathways through which both Stimulus 1 and Stimulus 2 propagate during learning, and 2) a set of adjacent spines on a neuron (for example, on a dendritic branch of neuron N5 where Stimulus 1 (cue stimulus) arrives) with the presupposition that synapses/ cluster of spines on the dendrite of a neuron have the capability for information storage. During memory retrieval, previous studies tried to correlate the changes at the synapses following learning with behavioral motor actions indicative of memory retrieval. But these studies did not search for a mechanism of generation of first-person internal sensations of memory. To understand its operational mechanism, our task is to use all the available information and examine the exact location where and how the learning-change is occurring that allows one of the stimuli (cue stimulus or here, Stimulus 1) to spark memory of the associatively learned second stimulus (Stimulus 2) in physiological time-scales of milliseconds along with provision for generating behavioral motor actions. Note that neuron N5 has two dendritic spines on its dendrite. Also note that the mean inter-spine distance between adjacent spines on a dendrite is more than the mean spine diameter. Our task is to discover where the interaction between pathways through which Stimulus 1 and Stimulus 2 propagate. In other words, we have to discover where and at what level the connections from neuron N9 will interact with the path through which Stimulus 1 propagates. Later arrival of Stimulus 1, the learning-induced interaction should be able to exhibit a unique mechanism to generate units of first-person inner sensations of memory of Stimulus 2. When the learning change remains only for a short period, it should be responsible for generating working memory. It is also expected to have the capability to get stabilized for long duration and is responsible for generating long-term memory. Inset: A synaptic junction between neurons N4 and N5 (marked Pre 1 and Post 1) along the path through which Stimulus 1 propagates is shown. The still unknown mechanism is shown as a large black box next to the synaptic junction. To decipher the secret, it is necessary to view memories as first-person inner sensations generated within milliseconds. Since learning can occur in milliseconds, it is also necessary to search for a learning mechanism that occurs in milliseconds from which memories can be generated. Semblance hypothesis has provided a solution for the contents in this black box that can explain how 1) the association between Stimulus 1 and Stimulus 2 is completed during learning within milliseconds, and 2) at a later time when the Stimulus 1 arrives (as a cue stimulus) how does it generate internal sensation of memory of Stimulus 2 as a first-person property within milliseconds, which can only be accessed by the (owner of the) system. It must also explain how the Stimulus 1 (cue stimulus) generates motor actions reminiscent of the arrival of Stimulus 2 as expected of a conditioning paradigm. In addition, it should provide explanations for a large number of features of the system observed at different levels. It is hoped that the readers will be able to find a testable mechanism by the end of reading this page FAQ pages of this website. This hypothesis has made several testable predictions.

There are several unsolved problems in neuroscience (Adolphs, 2015) observed from different levels (Edelman 2012; Gallistel & Matzel 2013). Nervous system functions observed at different levels are being studied by different faculties of science - biochemistry, cell biology, electrophysiology, systems neuroscience, psychology and consciousness studies. The system is similar to a puzzle lying in multiple dimensions. Solving it requires fitting  the correct pieces of the puzzle at the right levels to obtain the right operational function. If we examine only one or a few levels of the system, we might arrive at certain solutions that will allow fitting together pieces of the puzzle only for those few levels. Features of the unexamined levels will most likely remain unexplainable, and we won't reach a solution for the system. The diverse nature of findings from different levels strongly indicates that the solution is going to be a unique one. At the same time, it is also expected to be a simple one. To solve the system and find out the correct mechanism, it is necessary to examine representative functions from all the levels simultaneously.

   A second view of the problem can be described as follows. It begins by examining the following situation: The heart pumps the blood, and the kidneys filter the waste materials from the blood; these functions are observed by us from a third-person perspective. We understood their functions quite well, evidenced by our ability to develop their replacements, such as artificial heart and dialysis. What functions does the nervous system carry out? What operational mechanism do we need to understand from brain to replicate/replace it? Brain generates an inner sense of the external world during perception, stores sensory information during associative learning and later produces the internal sensation of retrieved memories of the learned item when the associatively learned cue stimulus arrives, induces thought process to connect different items from different sets of learning events – all of which are first-person properties that cannot be accessed by third-person observers. The only information that are available from the owner of the nervous system to a third-person are from the surrogate markers consisting of motor activity - specifically, speech and behavior. Therefore, the pieces of the puzzle mentioned in the above paragraph should be capable of explaining both first- and third-person accessible functions.


   A third view of the problem becomes evident when examined from the viewpoint of a builder. Here, the job is to replicate the nervous system in an engineered system. Since intentionality to feed and carry out all the actions for survival and reproduction are present even among the members of lower-level animal species, a robust evolutionarily conserved circuit mechanism for generating internal sensations is expected to be present in all the nervous systems. Since the first-person properties cannot be accessed by third person observers, they cannot be studied directly using biological systems. Hence, we should keep replication of the mechanism in engineered systems as the gold standard proof. These systems need to be built to provide readouts for the first-person internal sensations that can be accessed by the third-persons. As a builder, we will feel the pressure to know how the system can operate to generate different functions. We are forced to speculate all the possible mechanisms and figure out the correct one. We will be concerned mostly with explaining the formation of internal sensations in physiological timescales. Before building the system, we need to draw a sketch of the system operations.


A fourth view becomes possible by observing the “loss of function” states of the system occurring at various levels. This can help understand the nature of the pieces of the puzzle. Early genetics research has gained valuable information from "inborn errors" of metabolism that provided guidance to understand the allelic organization of genes. In this context, examining neurological and psychiatric disorders can help to understand the nature of the operating mechanism. Since the exact pathological features of many of these diseases are not yet known, it is expected that the loss of function of the operational units that induce both the first-person and third-person features is expected to provide information about the pathologies from which the function can be verified.


A large number of features observed by different subfields of neuroscience and psychology are required to be explained by a solution for the system. Since these features are very diverse, only a unique cellular mechanism will be able to explain all of them. This is expected to be a unique structure-function mechanism occurring at the intersection between the third person observed features and first-person properties. In other words, it is a dynamic, but stabilizable structural feature that can provide basic units of first-person internal sensations of different higher brain functions. It is necessary to verify whether the derived solution can explain findings from different specialized faculties within the large fields of neuroscience and psychology and test whether the explanations are inter-connectable. Since our current research efforts in each field are moving towards more specialized and super-specialized areas, finding and verifying the unique solution (to put the pieces of the puzzle together) requires an effort in the opposite direction. Anticipating this is the most important requirement for solving the system. It is necessary to explain how the nervous system functions occurring at different levels, such as - a mechanism that directs potentials to induce the internal sensation concurrent with the activation of motor neurons at physiological timescales (interconnecting central mechanism), dendritic spine changes, long-term potentiation (LTP), place cell firing, consolidation of memory, and association of memory with a feasible framework for consciousness - are interconnected.


For arriving at an operational mechanism that can explain both third person and first-person properties, a theoretical approach is the most efficient method. Among different brain functions, memory has the advantage that experiments can be carried out both to associatively teach the system and examine how these learning-induced changes are used during memory retrieval. A very large amount of experimental data is available in the field of memory research. Since no cellular changes are observed during memory retrieval, the memory retrieval is likely taking place by a passive reactivation of a learning-induced change. Memories were classified into working, short-term and long-term based on the differences in the period of time, following learning, during which they can be retrieved. Studies have been carried out with the assumption that the cellular mechanisms during learning that leads to memories classified in this manner are different. Since qualia (virtual first-person internal sensations) of these retrieved memories are almost same, it prompts one to ask, "What if a) a common cellular mechanism is taking place during learning, and b) the retrieval of different types of memories can be explained by reactivation of learning-induced changes that are retained for different durations?" To undertake such an experimental approach, one may ask, "Can we directly examine the memories themselves instead of examining the motor activity such as behavior and speech at the time of memory retrieval?" This will also eliminate examination of slow molecular changes following learning. In this context, it is necessary to re-define the question: "What are memories?" Memories are first-person virtual internal sensations of an item (in the absence of that item) induced within the nervous system (in response to a cue stimulus or occurring spontaneously). The sensation of a stimulus in its absence is hallucination. Therefore, memories can be viewed as cue-induced cue-specific hallucinations. Can we search for a learning-mechanism that can allow induction of virtual first-person internal sensations of memory as a cue-induced hallucination? This is the basis of developing semblance hypothesis which was published first as a book in 2007 (a copy is uploaded in Publications section). Revised editions were published in 2008 and 2010.

There are three main reasons why we had difficulties in solving the nervous system.

1) Frame of reference problem: Every time we had a frame of reference problem in the past, we needed to pause for some time to make further progress. A typical example is the difficulties in sensing the rotation of Earth (note that the speed of rotation of Earth on its axis is 1670 km per hour). Our sensory systems cannot sense rotation of Earth towards the East since we are in the same frame of reference as that of the Earth. The fact that third-person observers cannot sense first-person inner sensations in a subject can be viewed as a frame of reference problems. Physics has developed methods to solve frame of reference problem. In the case of the nervous system, we need to use the principles of methods used in physics to overcome this hurdle to  successfully solve it. Since this is new for neuroscience, there will be some difficulties in thinking about it in the beginning. But hopefully we will soon understand the problem and accept methods to overcome it.   

2) Difficulty to study the “virtual”: Another difficulty is the virtual nature of first-person inner sensations. However, we have had the experience of dealing with virtual items in the past. For example, numbers do not exist. We made them. In fact, they are virtual in nature. We can say that they represent real counts of items. What about negative numbers (integers?). They can exist only in our imagination. Yet, we use them routinely in mathematics. On a graph, we don’t feel their virtual nature at all. Going one step further, we have invented complex number (imaginary number). This solved our difficulties in finding square roots of negative integers, which helped to further advance mathematics. In a similar manner, once we understand where and how units of inner sensations are sparked within the system, we will be able to perceive a virtual space where we can position them and navigate that space to understand their different conformations.

3) Access problem: How to understand something that cannot be accessed by our sensory systems? There are several examples where we succeeded in overcoming such problems. For example, we cannot see DNA inside the cells or in a gel. But we stain it with ethidium bromide that will allow us to see the stain through our eyes. This means that our access problem can be overcome by adding one step that allows us to use our sensory system (here, our eyes). We sometimes go two steps. For example, we use a primary antibody to an antigen, followed by a secondary antibody with florescent property that can then be visualized through our eyes. Even though our sensory systems do not have direct access, we believe that an antigen is present based on logic that we use in our staining protocol. Understanding inner sensations will need another such indirect method that we will eventually become familiar with .

How did Galileo show us to put together observations to make an inference even when our sensory systems cannot agree with such an inference?

When are many observables are present in a system, putting them together can provide a totally new inference that our sensory systems may not be able to directly sense. So, naturally such type of an inference often will not be accepted by our community quickly. But every scientist has learned how to uphold the importance of this scientific method. An often-cited example is that of the inference made by Galileo Galilei using different observations that he made using his new telescope. Galileo published his findings in his book “The starry messenger.” While observing Jupiter on 7th January 1610, Galileo found four moons that are orbiting Jupiter. He immediately made the conclusion that if these moons are revolving around Jupiter, then it is unlikely for the Earth to be at the center of the Universe. A video explains this.

Galileo then turned his telescope towards Venus. It showed phases similar to that of Moon - New Moon and Full Moon. Galileo observed both the New and Full faces of Jupiter. When it is New, it is very big. When it is Full, it is very small. Galileo concluded that this could happen only if Jupiter revolves around the Sun and therefore, he made the inference that Sun must be at the center of the system. A suitable fit with this observation is that Earth and other planets are also revolving around the Sun. What we now know is that Jupiter is New when it comes close to the Earth, blocking the light from the Sun. When Jupiter is on the other side of the Sun during its revolution around the Sun, it is farthest from the Earth. Here, Jupiter is seen as small, and its face is Full. A nice video explains it is here.

The above two observations made Galileo conclude that Earth is not at the center of the solar system. Galileo saw the simplicity if all planets revolved around the Sun. Galileo was making logical arguments that allowed him to fit all the findings together. We have such a rich tradition in science to gather information from different observations and then put them together to make an inference even though we cannot appreciate it directly with our sensory systems. Another limitation of our sensory systems is their inability to sense the speed of rotation of the Earth on its axis. Since the circumference of the Earth at the equation is nearly 40,000 kilometers and one day has 24 hours, we can make the inference that the speed of rotation of the Earth is nearly 1650 kilometers per hour. Even though we cannot sense this speed using our sensory systems while on Earth, our inference about the speed of rotation of the Earth (and ours) must be true. When we observe the Earth from space, we see rotation of the Earth. See NASA’s video from the international space station (ISS) located 408 kilometers from Earth. (This time-lapse video from ISS needs to be converted to real-time video).

In short, we must find ways to overcome the limitations of our sensory systems! So, the question is, "How can we understand the operation of the nervous system even without replicating the mechanism in engineered systems?" In the case of the nervous system, our job is to put all the observations together to make an inference that will interconnect all those observations. It is obvious that the main function of the nervous system is generation of inner sensations as a first-person property, which a third person observer cannot sense directly. We have been thinking that it is the most difficult function to understand. However, we must obtain motivation from the above examples of finding solutions for non-sensible events. We must be willing to derive a testable solution using observations from multiple levels of the system and put forward testable predictions that we can verify.

We need to bring all the observations of the nervous system together and make an inference (solution). In this exercise, we cannot afford to leave out even a single observation since we must make sure that we reach a solution that can interconnect all the features of the system. When we become able to derive a solution that can provide explanations for all the functions, then we can make a reasonable assumption that the derived solution is correct. We must use this solution to make testable predictions that we can go and verify.


  It was Copernicus, who based on observations, first proposed that the Sun is stationary and that the Earth and other planets revolve around it. So, the credit must go to him as the first person who reported that Earth is revolving around the Sun. Even though Gallio thought that planets are moving in circles, Kepler found about the elliptic path of motion of planets.

How can we solve the nervous system?

As we face situations that have more steps away from reality, we have to rely on our logical reasoning capabilities (see an excerpt from Krakauer et al., 2017 that appeared in the journal Neuron). We have a very large number of findings from different levels, and we have to discover the solution that can interconnect all of them. We need to overcome a) the frame of reference problem, b) the access problem, and c) to deal with the virtual nature of internal sensations. But what if we cannot sense one part of the solution directly by our sensory systems while trying to find the solution? In this case, it is possible to seek examples of approaches that are used by other fields of sciences. For example, physics study particles and fields that are not accessible to our sensory systems. What is the deep underlying principle behind their success? A summary is given in Table 1 below.



1) First, a large number of observations are made that appear to be disparate in nature. This means that these findings cannot be explained in terms of each other. 1) There are a large number of disparate findings in neuroscience (see Table 2) that need inter-connectable explanations. Example: How does the operation of the system relate to sleep and also to the electrophysiological finding of LTP?
2) The above indicates the presence of a deep underlying principle that should interconnect these disparate observations. 2) There should be a deep underlying principle behind all those observations in Table 2.
3) The effects of the above principle are the ones (e.g. particles and fields) that cannot be directly sensed by our sensory systems. 3) There is a principle, the products of which (inner sensations) cannot be sensed by our (third-persons') sensory systems. Yet, the principle of the mechanism should be able to explain and interconnect all the observed findings.
4) The next step is to search for any possible solution that can interconnect all the findings. Constraints provided by disparate observations should be able to guide us towards the solution. This is done either by initial deduction followed by mathematical approximations (Special and General Relativity) or by pure mathematical derivation (Higgs Bosons). 4) A structure-function mechanism has to be sought by logical deduction & trial and error methods. All the constraints offered by a large number of findings can be used to derive the solution. Success depends on moving through a path as guided by all the constraints. Only when we reach the correct solution, we will be able to explain all the findings in an interconnectable manner.
5) The solution is then confirmed by verifying the testable predictions. 5) Testable predictions made by the derived mechanism can be verified.

Table 1. Comparison between the steps taken by physics and neuroscience when it becomes necessary to unify disparate findings. This is important especially when dealing with properties that are non-accessible to our sensory systems such as particles, fields, and first-person internal sensations etc.

The deep underlying principle of many studies in physics has similarities to the method used in linear algebra for solving a system of large set of linear equations that has a unique solution. If one tries to solve such a system, one can find that the relationships between the variables in each equation provide hints that can guide us towards the solution. If there are a large number of variables, there should be at least an equal number of equations to find the unique solution. Since there is a large number of findings at different levels of the nervous system, we can (and we must) use even minor interconnected features between them to find the solution. It is a gigantic exercise since there are no easy methods in biology like that are used in linear algebra. But understanding linear algebraic methods will definitely help in solving the system.

In linear algebra, Gauss-Jordan elimination method is used to find the solution for a system of linear equations. If we look carefully, we can see that easy methods in linear algebra were designed by someone who understood the deep underlying principle and worked on to make it simple for others by developing easy methods. We can examine how the relationships between variables in each equation define the unique solution for a system of linear equations and how Gauss-Jordan elimination method was invented. It is to be noted that we can also solve linear algebra problems using trial and error methods. But it will take some extra time. In other words, in mathematics easy methods are developed for convenience. Whichever method is used, the deep underlying principle is the same - A system exhibiting a large number of disparate findings (equations) most likely has a unique solution that binds (interconnects) all the findings (equations) within that system. By finding a solution that can interconnect a subset of findings and by repeating this approach using different subsets of findings, one can hope to reach a common underlying solution, which is the correct solution. One can start attempting to solve the (nervous) system by using subsets of disparate findings. The optimism with this approach is that there is only one unique solution for the nervous system, and it is easy to rule out many wrong solutions quickly. Using the above principle, subsets of constraints provided by findings from various levels (Table 2) were used to derive a mechanism that can explain, and interconnect findings made by different faculties of brain research. This approach is expected to lead to the derivation of an operational mechanism for the generation of internal sensations, of which the non-sensible component will continue to remain non-sensible to our sensory systems even after its discovery (Figure 2). However, it is expected to show testable learning-generated change that gets reactivated at the time of memory retrieval to induce basic units of internal sensation whose integrational product can provide sensory qualia of the retrieved memory. Structural and electrophysiological changes that are expected to occur from these changes are expected to get explained using experimental results from different laboratories.  

Deriving solution of the nervous system

Figure 2. Method to find a solution for the system from third person observed findings. (A) Features of the system are sensed either directly (represented by capital letter K) by our sensory systems or indirectly (represented by capital letters D, E, G, H) through findings such as staining of proteins, observing behavior, etc. (represented by small letters (u, w, v, x) connecting the features through straight lines (for example, observation of u enables sensing D). (B) Using both commonly used direct and indirect methods, three clusters of interconnected (represented by dotted lines) findings are found at separate levels (observations from different fields of brain science). In most cases, it was not possible to interconnect these clusters. For example, it was not possible to find interconnected explanations between 1) learning changes and inner sensation of memory both occurring in millisecond timescales, and 2) sleep and LTP. Using constraints from findings within each cluster, it is possible to examine whether they can be interconnected through a common operational mechanism. In the case of the nervous system, very large number of findings and constraints offered by them can be examined. (C) Using constraints available from some of the features of each cluster of unrelated findings (e.g. A, B and C), it is necessary to try to derive a deep underlying principle (a structure-function solution m) that allows interconnection between them and therefore all the findings within each cluster. This solution is expected to provide a mechanism for generation of internal sensations in millisecond timescales. (D) The solution m enables explaining how various findings within each cluster are interrelated with each other and with the findings from other clusters shown in B). While remaining non-sensible to our senses by any known methods used in current biological investigations, ability of the solution m to hold different findings from all the clusters together makes it a further verifiable solution (Figure from Vadakkan KI (2019) From Cells to Sensations: A Window to the Physics of Mind. Physics of Life Reviews. 31:44-78).

There are very large number of observations from different levels of the nervous system functions (Table 2). An example of store receipts (this is also the deep principle behind some of the approaches carried out in physical sciences to understand Nature) provides us confirmatory evidence that we can reach a final correct solution, if it becomes possible to obtain large number of findings that contain all the variables of the system. From a philosophical standpoint (Excerpts from an article here) also, we need to undertake a similar approach by examining findings from all levels together. This approach is expected to provide information about a testable mechanism that generates first-person inner sensations.


Nonsensible (not directly with our senses) features of a system can be inferred from its sensible properties. Since we have a large number of constraints from findings in multiple levels of the nervous system, a derived solution that can explain all its features at those levels is likely correct even if we are unable to directly sense the formation of first-person inner sensation. Further, we can a) do retrodictive examinations, & b) make testable predictions for verification.  Pdf of these constraints here: Pdf



The findings.

Constraints offered by the findings (on the left side) that direct the enquiry towards a correct solution.

Interconnected explanations by the semblance hypothesis (Please read this row after reading the hypothesis).


Both associatively learned stimuli & prompt (cue) stimuli propagate through synaptically-connected neuronal circuit.

Mechanism should operate synchronous with the synaptically-connected circuitry.

Inter-neuronal inter-postsynaptic functional LINKs (IPLs) form and operate only when synaptic transmission takes place (Vadakkan, 2007; 2013).


Learning-induced changes occur in physiological timescales (in milliseconds)Foot note1

A learning-inducible change that occurs (& completed) in physiological timescales (to explain the ability to retrieve memory instantly following learning).

Propagation of potentials along the IPLs to the inter-LINKed spines takes place in physiological timescales of milliseconds (Vadakkan, 2007; 2013).


Memories that can be retrieved after long period after learning are also capable of getting retrieved immediately following learning (working memory).

Learning should generate retrieval-efficient changes within milliseconds. They can be used for memory retrieval immediately (working memory). These changes have a provision for remaining in a stable form for long period, responsible for long-term memory.

IPL formation takes place at the time of learning. When IPL reverses back, then its short duration can only generate working memory. The life span of IPL decides the duration of storage of memory. Long-term stabilization of IPLs lead to long-term memory (Vadakkan, 2010a; 2013).


When exposed to a cue stimulus, inner sensation of memory occurs in  physiological timescales (in milliseconds).

A learning-induced change should be capable of inducing inner sensation of memory in physiological timescales (completed within this time).

Propagation of potentials along the IPL generates semblance over the inter-LINKed spine instantly as a system property (Vadakkan, 2013).


Memory is an internal sensation of an item/event having certain specific sensory features (qualia).

Mechanism is expected to have elements that can provide sensory features to the retrieved memory.

Integration of all the units of inner sensation generated on large number of inter-LINKed spines by a cue stimulus provide qualia (Vadakkan, 2010a).


Ability to store large set of learning-induced changes responsible for retrieving large number of memories.

Neurons and their processes are finite in number. Therefore, an efficient operation for storing large numbers of learning-induced changes becomes possible if common elements in each learning mechanism can be shared. Hence, each memory is expected to get induced from a combination of unitary mechanisms. 

Inter-LINKed spines within the islets of inter-LINKed spines can be depolarized by any specific stimulus reaching it. Since items/events consist of combinations of sensory stimuli, different combinations of inter-LINKed spines can be used by different cue stimuli to generate corresponding memories (Vadakkan, 2010a).


Instant access to very large memory stores.

A specific cue stimulus should be able to induce a specific memory by combinatorial reactivation of a specific set of learning-induced unitary changes.

Inter-LINKed spines can be accessed by stimuli from any cue stimulus (Vadakkan, 2010a).


Absence of cellular changes during memory retrieval.

A passive reactivation of the changes that occur during learning should be getting used at the time of memory retrieval to induce units of internal sensations. This should take place at physiological timescales of milliseconds. 

Since inter-LINKed spines persist from the time of learning, propagatoin of depolarization along the IPL does not require any new cellular changes (Vadakkan, 2010a).


During memory retrieval, firing of a subset of neurons that were not firing before learning in response to the same cue stimulus occurs.

Learning has opened certain new channels & cue stimulus leads to propagation of depolarization through these channels to provide additional potentials to a subset of neurons that are otherwise being held at subthreshold activation state (without firing). This will lead to firing of these set of neurons.

Formation of IPLs during learning will lead to propagation of depolarization across them to the inter-LINKed spines. This will provide additional potentials to the inter-LINKed spine’s neurons and may fire those neurons (Vadakkan, 2013).


Brain operates in a narrow range of frequency of extracellularly recorded oscillating potentials.

Expected mechanism provides vector components of the oscillating potentials.

Synaptic transmission & near perpendicular propagation of depolarization across the IPL provides vector components. Reactivation of a set of IPLs at rest generates a background semblance, which is expected to generate inner sensation of a conscious state (Vadakkan, 2010b).


Motivation promotes learning. Motivation is associated with release of dopamine at different locations of the brain.

Motivation is associated with specific factors and their specific actions are expected to promote the learning-induced change and possibly to retain this change for longer period than that occur in its absence.

Dopamine is known to cause spine expansion (Yagishita et al., 2014). Expanding spines can augment IPL formation and retain the formed IPLs for long time that may trigger some stabilization steps.


Internal sensations of working, short, and long-term memories have similar qualia.

The same learning-induced change is retained for different durations. Long-term memory loses its clarity both due to loss of some unitary mechanisms & dilution of specificity by combining with newly formed units of inner sensations.

Memories of all durations take place by reactivation of inter-LINKed spines by depolarization propagating along the IPLs and generating units of inner sensations (Vadakkan, 2010a, 2013).


Working memory lasts only for a very short period of time.

Learning-induced change must have a quickly reversible mechanism.

IPL formation is a high energy requiring process as can be inferred from experiments using artificial membranes (Rand and Parsegian, 1984; Martens and McMahon, 2008; Harrison, 2015). Hence majority of IPLs are expected to reverse quickly.


Some of the same memories that are retrieved as working memories can be retrieved after a long period of time after the learning.

Learning-induced change must be able to undergo certain changes that will enable it to get maintained for a long period.

Stabilization of IPLs for long period can induce the same units of inner sensation. If the number of IPLs that can be reactivated by a cue stimulus after a long period decreases, then qualia of memory will reduce (Vadakkan, 2010a; 2013).


Simultaneous existence of previous two conditions (above two rows) within the system.

Learning-induced mechanism should have an initial quickly reversible change that under certain circumstances can progress towards a stage where it can get stabilized for long period of time.

When memory of a beneficial or deleterious item or event becomes advantageous for survival, then IPLs necessary for those memories are stabilized for long period (Vadakkan, 2010a; 2013).


Internal sensation of memory in response to a cue stimulus varies with the nature of the cue stimulus.

Specific sensory features from the cue stimulus induce a specific combination of internal sensory units to generate specific features of memory being retrieved.

As the cue stimulus propagates through its path, depolarization propagates through a specific set of IPLs that induce a specific set of units of inner sensations for a specific memory (Vadakkan, 2010a, 2013).


After associative learning between two items, arrival of one of the items generates memory of the other item.

The learning mechanism should have features to explain how either one of the associatively learned items can act as a cue stimulus to generate memory of the other item. Hence, the mechanism should have the ability to show bidirectionality in it.

Semblance can be generated from either one of the inter-LINKed spines on the two sides of an IPL (Vadakkan, 2010a; 2013).


Even partial features of one of the associatively learned item is capable of retrieving memory of the second item.

The mechanism should have features to explain how stimuli from partial features of one stimulus can retrieve memory of the second item.

Partial stimuli propagate to generate inner sensation of a framework of an item or event. Due to the property of generalization (by virtue of spread of depolarization across the entire islet of inter-LINKed spines), it is expected to provide more features to the memory (Vadakkan, 2010a; 2013; 2019a).


Ability to store new memories without needing to overwrite the old ones

Sharing of unitary mechanism for common features, and provision for formation of new units with new associations are expected to be present in the system

Inter-LINKed spines within islets of inter-LINKed spines can be shared by any stimuli reaching them. Hence there is no need for overwriting old memories. Any few associations can get inter-LINKed to the existing islets of inter-LINKed spines (Vadakkan, 2010a; 2013).


Consolidation of memory (Transfer of storage locations of memory from the hippocampus to the cortex) over a span of 5 to 8 years.

Addition of specific learning-induced changes in the cortex over time using similar unitary sensory associations, and ability to generate memories by a global integrating mechanism. Must go through a stage of surplus unitary mechanisms.

Convergence of all sensory stimuli in the hippocampus leads to dense islets where inter-LINKed spines can be formed. Sparser IPLs are expected to be formed in the cortex. Neurogenesis and repetition of associative elements within each learning will lead to formation of surplus IPLs in the cortex over time (Vadakkan, 2011a).


Mechanism uses pre-existing schemas (Tse et al., 2007). Schemas are expected to get used inter-changeably.

Changes induced by one learning are shared by another learning event. For this to occur, there must be shared unitary mechanisms in each learning event and presumably also in the memory retrieval mechanism.

Inter-LINKed spines can be used by any cue stimuli, allowing the unitary structural operations to get shared. It is also reasonable to assume that units of inner sensations induced from these inter-LINKed spines also get shared by similar elements in stimuli (Vadakkan, 2010a; 2013).


A constantly adapting dynamic circuit mechanism is expected.

Provisions should be present to accommodate large number of new learning events.

Since extracellular matrix space is minimal in the cortex, large number of spines from different neurons are expected to remain abutted to each other. They can be readily inter-LINKed depending on the nature of cue stimuli (Vadakkan, 2010a; 2013).


Framework of a mechanism that can generate hypothesis by the system.

When one of the elementary mechanisms of one associative learning event undergo association with a third elementary mechanism during a third associative learning, it will lead to an interconnected chain of associations (ICAs). When there are common elements in different ICAs, then the system will be able to generate a hypothesis of relationships between events/items.

When one spine each from two islets of inter-LINKed spines are inter-LINKed, any one spine from one islet establishes a relationship with any one spine in the second islet of inter-LINKed spines. When more than one islet of inter-LINKed spines is inter-LINKed in this manner, it leads to generation of hypothesis about something that will not occur ordinarily (Vadakkan, 2010a; 2013).


System needs a state of sleep for nearly one third of its operational time.

It is necessary to explain why the system won't be able to exist without sleep. i.e. Explain substantive nature of sleep in the operation of the system.

State of sleep is needed to keep postsynaptic depolarization by presynaptic terminal as a dominant state of the system. Only when this dominant state is maintained, then only a lateral activation of the postsynaptic terminal (spine) will induce units of inner sensation of a stimulus arriving from the environment through the presynaptic terminal (Vadakkan, 2016b). Since nothing comes from the environment, it is a hallucination (Minsky, 1980).


While living in space, requirement of sleep reduces by more than one hour.

Provide a mechanistic explanation why reduced sensory stimuli in space reduces the need for sleep. 

Since sensory stimuli is less in space, number of reactivations of inter-LINKed spines is reduced. This reduces the time to set the system to its baseline dominant state of postsynaptic depolarization by its presynaptic terminal (Vadakkan, 2016b).


During memory retrieval, inner sensation of memory can occur with or without motor actions such as speech or behavioral motor actions.

The mechanism that generates inner sensation of memory should have a connection with the mechanism that generates motor action. There should be a provision for disabling this connection at will.

IPL mechanism can generate both units of first-person inner sensations and motor action reminiscent of arrival of the item whose memory is retrieved. The motor outputs can be inhibited while inner sensation is being generated (Vadakkan, 2010a; 2013).


It is difficult to inhibit a memory which is being retrieved.

A structural mechanistic explanation is needed.

IPL is an inter-membrane connection. Once IPL is present and functions, it is not possible to inhibit its function voluntarily. Additional inter-spine LINK with an inhibitory spine may become possible through future associative learning events (Vadakkan, 2007; 2010a).


Mean inter-spine distance on the dendrite of a pyramidal neuron is more than the mean spine diameter (Konur et al., 2003).

This opens the possibility for neuronal processes that belong to other neurons to occupy the inter-spine space. It is reasonable to expect some functional importance for such a scheme of inter-spine spacing. Since spines of different neurons occupy this space and ECM is often negligible (see Fig.13 in FAQ), some of the spines that belong to different neurons can remain abutted to each other.

Abutted spines that belong to different neurons increase the probability for inter-neuronal inter-spine interactions. These interactions are the basis of IPL proposed by the semblance hypothesis (Vadakkan, 2010a; 2013).


Both learning and retrieval of memory are associated with firing of a set of neurons.

Both learning and memory retrieval allow potentials to propagate through certain gateways that allow certain neurons that are being held at sub-threshold activation states to fire action potentials. This gateway may be formed at the time of learning and are likely associated with generation of units of inner sensation of memory.

Both learning and memory retrieval allow propagation of potentials along the IPL. This leads to the arrival of more potentials to the axon hillock of the postsynaptic neuron. If arrival of these additional potentials can allow the subthreshold activation state of the neuron to cross the threshold for firing, it fires an action potential (Vadakkan, 2010a; 2013).


Place cells fire in response to specific spatial stimulus.

Mechanism that generates inner sensation of memory for a location is expected to have a mechanistic connection with firing of a set of CA1 neurons.

Place cells are CA1 pyramidal cells. When islets of inter-LINKed spines of overlapping dendrites of CA1 neurons receive spatial inputs, they provide potentials to their postsynaptic CA1 neurons. If these CA1 neurons are being held at subthreshold activation states, then they fire. This explains place cell firing (Vadakkan, 2013; 2016a).


Firing of an ensemble of neurons during a higher brain function.

Inner sensation generated during a higher brain function is related with firing of an ensemble of neurons.

Reactivation of IPLs during a higher brain function will add potentials from the inter-LINKed spines to their postsynaptic neurons. If these potentials add up to allow these neurons to cross the threshold, they will fire (Vadakkan, 2010a; 2016a).


Firing of separate sets of neurons during learning and memory retrieval.

Learning and generation of inner sensation of memory are associated with firing of separate sets of neurons.

When memory retrieval is immediately following learning, then lack of stimuli from the item whose memory is being retrieved is responsible for the difference. When memory is retrieved late, other associative learning events in between the learning and memory retrieval under examination will generate additional IPLs in the circuitry & will be responsible for the difference (Vadakkan, 2010a; 2016a).


Fast changes in both the magnitude and correlational structure of cortical network activity (Benisty et al., 2024).

Rapidly time-varying functional connectivity is responsible for such changes.

Changes in environmental stimuli, self-triggered thought processes, inner sensations of fear, anticipation, hunger, and comfort levels fluctuate moment to moment indicating reactivation of a new set of IPLs. This will change network activity (Vadakkan, 2019a).


Firing of a cortical neuron (axonal spike) is possible by summation of nearly 140 postsynaptic potentials (input signals) arriving from random locations. Each of these cortical neurons have tens of thousands of dendritic spines where postsynaptic potentials get generated.

These neurons must be maintained at a sub-threshold state in the background state and the mechanism of induction of internal sensation must be associated with providing additional postsynaptic potentials for crossing the threshold for firing of these neurons.

Recent modelling studies have shown that a pyramidal neuron that has tens of thousands of input connections can fire an action potential by spatial summation (summation at the same time) of nearly 140 EPSPs at the axonal hillock that arrives from randomly located dendritic spines (Palmer et al. 2014; Eyal et al., 2018). Based on calculations of energy per bit of information 2000 synaptic inputs are needed for neuronal firing (Levy and Calvert, 2021).


Any set of 140 input signals arriving from random locations on the dendritic tree can fire a neuron. Hence, there is extreme degeneracy of input signals in firing a neuron. A system operating by such a scheme was selected from large number of variations since this was offering functional advantage to the system.

Since such a scheme is expected to be used specifically, then a possible situation must be there. If a neuron is being held at subthreshold level by receiving nearly 130 inputs, then it needs 10 more input signals for its firing. If only a specific cue stimulus in a specific context can provide a specific set of input signals for 10 additional input signals, then this possibility can be tested.

Islet of inter-LINKed spines can provide an opportunity to pool all the potentials at one location from where it can be delivered in a summated manner. Dendritic spikes can be viewed resulting from it. These can reach the axonal hillock efficiently to cause neuronal firing for motor effect. Inter-LINKing with spines that receive different neurotransmitters at the islets can regulate these islets (Vadakkan, 2016a).


Many neurons are being held at sub-threshold activation state.

By holding a neuron at a certain potential below the threshold, it is possible to regulate the neuronal output conditional upon arrival of certain number of inputs. If these inputs can be made conditions for an output, then it has operational significance

Several neurons are being held at subthreshold activation states (Seong et al., 2014). At the islets of inter-LINKed spines summation of potentials from certain set of inputs can be guided to generate summated potentials and even spikes that can decide motor outputs



Input signals (postsynaptic potentials) have maximum strength at the location of their origin, which is the spine head. As the potentials propagate, they get attenuated in the spine neck region. Further attenuation occurs as they propagate towards the neuronal cell body.

When signals from a stimulus attenuate, they may not contribute to an efficient learning mechanism. Furthermore, signals from different spines mix in the dendrite. Hence, most likely location for a learning mechanism that can maintain specificity until the time of its retrieval is expected to occur at the spine head region.

IPL occurs between spine head regions between spines that belong to different neurons (Vadakkan, 2010a; 2016a).


Dendritic spikes occur by the summation of nearly 10 to 50 postsynaptic potentials (of the spines) at the dendritic region.

It is necessary to explain which spines contribute to the potentials and explain their significance.

Though not proved, semblance hypothesis proposed that most of the potentials that contribute to dendritic spikes originate from spines that belong to different neurons that for islets of inter-LINKed spines (Vadakkan, 2016a).


When current is injected into the dendrites of human layer 2/3 neurons they generated repetitive trains of fast dendritic calcium spikes, which can be independent of somatic action potentials (Gidon et al., 2020).

Explanation for spike is needed.

The islets of inter-LINKed spines can lead to generation of dendritic spikes. The net potential can drain through some of the spines depending on the several regulatory factors (Vadakkan, 2016a).

40 Certain dendritic spikes are not followed by somatic action potentials (Golding & Spruston, 1998) Conventionally it is thought that dendritic spikes are efficient detectors of specific input patterns ensuring a neuronal output (action potential) (Gasparini et al., 2004). So, a source for leakage of potentials from the dendritic area other than its propagation towards the soma needs to be explained. The islet of inter-LINKed spines (IILSs) provides a new route for the dendritic spike to propagate. Dendritic spike can backpropagate to many IILSs & propagate through their inter-LINKed spines (that offer less resistant routes) towards the dendritic trees of different neurons from which we are not recording (so we don't record the leak).


Inner sensation of certain higher brain functions occurs without any motor actions.

Either the motor action can be voluntarily suppressed or that there are no behavioral motor actions associated with it.

The apical dendrites in human layer 5 neurons are electrically isolated from that of the somatic compartment, possibly having independent operations of islets of inter-LINKed spines at those remote dendritic regions (Beaulieu-Laroche et al., 2018).


When two differential electrodes are placed at 2 extracellular locations, extracellular potentials can be recorded. They show oscillations. Brain operates only when the frequency of these oscillations occurs within a narrow range.

While synaptic transmission provides one vector component, something else constitutes the other vector component/s that is/are expected to take place nearly perpendicular to the direction of synaptic transmission. Brain functions (both first-person & motor actions) are linked tightly to these vector components.

Propagation of depolarization along the IPL provides a vector component almost perpendicular to that of the synaptic transmission (Vadakkan, 2010a; 2013).


Apical tuft regions of neurons of all the cortical neuronal orders are anchored to the inner pial surface resulting in overlapping of the dendritic arbors of neurons from different orders. This resulted from a sequence of movement of neuronal precursors during development.

Dendritic spines of neurons that belong to both the same (mainly) and different neuronal orders overlap with each other to serve certain functions.

Overlapping of dendrites that belong to different neurons facilitate formation of inter-neuronal inter-spine LINKs. Anchoring of apical tuft regions of all the cortical neuronal orders facilitates this (Vadakkan, 2016a, 2019a).


Following learning, initially there is conscious retrieval of memory and eventually this becomes sub-conscious after repeated retrievals.

The process by which repeated retrieval of a memory to a sub-conscious level must be able to explain a framework of a mechanism of consciousness.

Explained by the semblance hypothesis (Vadakkan, 2010b; 2019a). Once the retrieval of memories of a certain item becomes a routine, the inner sensations evoked by their IPL reactivations will be added on to the inner sensation of consciousness. Hence, such inner sensations will not be notice individually.


Experimental finding of long-term potentiation (LTP) has shown several correlations with behavioral motor actions that are surrogate markers of memory retrieval.

It must be possible to explain how cellular changes during LTP induction and learning are correlated & how this is related to the ability for memory retrieval.

Explained by the semblance hypothesis (Vadakkan, 2019b) Foot note2


Learning takes place in milliseconds, whereas LTP induction takes at least 20 to 30 seconds and even more a minute.

Cellular changes during learning are expected to get scaled-up during LTP induction in a time-dependent manner. Need a cellular explanation for this.

Explained by the semblance hypothesis (Vadakkan, 2019b).


Blockers of membrane fusion blocks LTP.

Need to explain the cellular location where they act and explain how it blocks LTP.         

Explained by the semblance hypothesis (Vadakkan, 2019b).


CA2 area of hippocampus is resistant to LTP induction. Induction of LTP here becomes possible by the removal of the peri-neural net proteins chemically.

Cellular mechanism responsible for LTP induction must be able to explain this. Perineural area is involved in the cellular mechanism of LTP induction.

Explained by the semblance hypothesis (Vadakkan, 2019b).


Several seizures spread laterally to adjacent cortices.

Cellular mechanism responsible for seizures should be capable of spreading laterally.

Explained by the semblance hypothesis (Vadakkan, 2016d).


Several seizures are associated with hallucinations.

Laterally spreading seizure mechanism should be able to explain how internal sensation of certain stimuli occur

Explained by the semblance hypothesis (Vadakkan, 2016d).


Amyotrophic lateral sclerosis (ALS) pathology spreads laterally.

Some structural aspects of the normal operational mechanism aid in the lateral spread of neurodegenerative changes under pathological conditions.

Explained by the semblance hypothesis (Vadakkan, 2016c).


Relationship between LTP, kindling, and seizures.

It is reasonable to infer that a structure-function-pathology relationship exists that can provide interconnecting explanations.

Explained by the semblance hypothesis (Vadakkan, 2016d).


Transfer of injected dye from one CA1 neuron to the neighboring CA1 neurons is observed in animal models of seizures (Colling et al., Brain Res. 1996).

Need an explanation for a physical connection between two CA1 neurons through which dye can diffuse.

Explained by the semblance hypothesis (Vadakkan, 2016d).


Loss of dendritic spines after kindling.

An interconnected explanation for kindling, seizures and LTP must also be able to be extended to explain loss of spines after kindling.

Explained by the semblance hypothesis (Vadakkan, 2016d).


CA2 area of hippocampus is resistant to seizures.

In row 44, we saw that CA2 region is resistant to LTP induction. Hence, whatever causes resistance to LTP induction must also be causing resistance to seizure generation.

Explained by the semblance hypothesis (Vadakkan, 2016d).


Seizures and memory loss are caused by herpes simplex viral (HSV) encephalitis.

Mechanistic explanation for both these features is expected to provide some information about the relationship between these findings in HSV encephalitis.

Explained by the semblance hypothesis (Vadakkan, 2016d).


Anesthetic agents alleviate seizures.

Mechanism of action of anesthetic agents should be able to explain how seizure generation and propagation are stopped by anesthetic agents.

Explained by the semblance hypothesis (Vadakkan, 2016d).


Cognitive impairment in patients with seizure disorders.

Mechanism of learning, memory retrieval and behavioral motor actions are expected to be affected by the mechanism of seizures.

Explained by the semblance hypothesis (Vadakkan, 2016d).


Intracellular electrophysiological correlate of epileptiform activity is paroxysmal depolarizing shift (PDS), which is a giant excitatory postsynaptic potential (EPSP).

A mechanistic explanation is needed for generation of a giant EPSP at the dendritic spine area during a seizure. It has a propensity to propagate laterally to other cortical regions. Need a mechanistic explanation.

Explained by the semblance hypothesis (Vadakkan, 2016d).


Neurodegenerative disorders show loss of spines and neuronal death.

An explanation is needed for contiguous spread of pathology leading to spine loss and neuronal death. Causative factors should be acting at specific locations to explain all its features.

Explained by the semblance hypothesis (Vadakkan, 2016c).


Dementia in neurodegenerative disorders.

Need an explanation for the role of spines in both generation of inner sensation of memory along with concurrent behavioral motor activity.               

Explained by the semblance hypothesis (Vadakkan, 2016c).


Perception as a first-person inner sensation.

A variant or a modification of the mechanism of induction of inner sensation for memory should be able to explain first-person inner sensation of perception.

Explained by the semblance hypothesis (Vadakkan, 2015b).


Apparent location of the percept different than its actual location.

Matching explanations using the mechanism of induction of units of inner sensation are needed.

The inner sensation of percept is generated by integral of all the perceptons. Hence, the actual location of an object need not necessarily match the percept. This becomes clear when there is a medium that shift the patch of light towards the eye (Vadakkan, 2015b).


Homogeneity in the percept for stimuli above the flicker fusion frequency.

A mechanism for fusion of inner sensation of continuous perception of a source of light that is affected by frequency of flickers is needed.

Since perceptons from IPLs located at different regions in response to one flicker has a temporal pattern of generation, overlapping formation of perceptons from consecutive flickers overlap and generate a continuous percept (Vadakkan, 2015b).


Perception of object borders. 

A mechanistic explanation for the formation of first-person percept for object borders is needed.

Only stimuli from within the border region reaches the brain. When perceptons formed from these stimuli integrate, they generate inner sensation of percept to generate boarder. Similarly, stimuli from outside the borders also do the same to generate a contrasting border of the background (Vadakkan, 2015b).


First-person inner sensation of pressure phosphenes.

Mechanism of generation of first-person inner sensations is expected to provide an explanation for phosphenes triggered by pressure over the eyeball.

Stimulation of sensory paths anywhere along it before reaching the locations of their convergence can lead to reactivation of IPLs for generation of perceptons (Vadakkan, 2015b).


Orientation tuning of a population of neurons in V1 before and after training on a visuomotor task showed different sets of neurons responding (Failor et al., 2021).

Neurons that fire as a consequence of associative learning changes in the primary visual cortex varies with time.

Based on the semblance hypothesis, the primary mechanism of perception is not through the firing of a specific set of visual cortical neurons. Instead, perceptons are generated at the inter-LINKed spines on either side of an IPL (Vadakkan, 2015b).


Flash-lag effect - When a flash is briefly presented in a specific location adjacent to the path of a uniformly moving object, the former is perceived to lag the latter.

Matching explanation using the mechanism of induction of units of inner sensation is needed. Needs to explain how perception is affected by relative time of arrival of a stimulus.

Explained based on the semblance hypothesis (Vadakkan, 2022). Visual pathway has synapses that cause synaptic delay. Overlapping reactivation of IPLs by continuous arrival of stimuli maintains perception; whereas a fresh stimulus undergo delay to initiate perception.


Inner sensation of consciousness.

Presence of a continuous operational mechanism for the generation of inner sensations that depends on/contributes to maintaining the frequency of oscillating extracellular potentials in a narrow range. The combined inner sensation is expected to generate inner sense of being conscious.

There is a baseline oscillating extracellular potentials as recorded by EEG. This shows testable propagation of potentials along many IPLs contributing to its horizontal component. Net inner sensation generated by reactivation of inter-LINKed spines during background state can contribute to inner sensation of consciousness (Vadakkan, 2010b).


Loss of consciousness by anesthetic agents.

Properties of anesthetic agents should be able to explain how the proposed mechanism of consciousness can be altered.

Explained by rapid chain formation of large number of non-specific IPLs (Vadakkan, 2015a).


Loss of consciousness during a generalized seizure and its reversal after seizure.

Mechanism of seizure generation should be able to explain how inner sensation of consciousness is lost.

Explained based on the semblance hypothesis (Vadakkan, 2016d). Rapid chain formation of large number of non-specific IPLs due to changes in ECM properties (e.g. Very low serum Na) or due to increased excitability of neurons.


Changes in consciousness with variations in the frequency of oscillating extracellular potentials beyond a narrow range.

Need an explanation how a narrow range of frequency of oscillating extracellular potentials is associated with normal state of normal state of consciousness.

Explained based on the semblance hypothesis (Vadakkan 2010b; 2015a). Unconscious states show large variations in the frequencies of extracellular potentials recordered from skull surface in EEG (Rusalova, 2006).


Effect of dopamine in augmenting anesthetic action.

Explain a mechanism how dopamine augments anesthetic action. This explanation must match with the explanation for the action of dopamine in augmenting learning (see row 11).

Explained based on the semblance hypothesis (Vadakkan, 2015a). Since dopamine can cause spine expansion (Yagishita et al., 2014), it will augment non-specific IPL formation by anesthetic agents.


Phantom sensation or pain.

Explain a mechanism for the inner sensation of pain from a lost limb at the time of phantom sensation or phantom pain.

As long as the IPLs that have received inputs from a limb remains stable in the brain, any reactivation of this by stimuli arriving to this IPL through a different sensory input can evoke semblance of phantom limb or pain.


Referred pain.

Explain a mechanism for the inner sensation of pain from a location different from the location where the cause of pain is present.

Inputs two different locations converge into one IPL at a higher neuronal order region can lead to semblance of sensory input towards those regions (Vadakkan, 2010a, 2013).


Mechanism for innate behavior that enables survival.

A mechanism evolving from heritable changes to explain innate behavior in response to a stimulus.

Explained based on the semblance hypothesis (Vadakkan, 2020). Convergence of sensory stimuli having different velocities is programmed in the genetic code and executed during development that favor the formation of IPLs.


Presence of a comparable circuitry in a remote animal species explains universal nature of a biological mechanism.

Comparable features that show relationship of a mechanism that induces units of inner sensation using synaptically-connected neuronal circuitry among different species of animals.

Organization of neuronal connections suggesting the presence of a comparable IPL circuitry in Drosophila olfactory system is explained (Vadakkan, 2015b).


Neurodegeneration resulting from repeated general anesthesia (Baranov et al., 2009).

Need an explanation why the repeated induction of a mechanism of loss of consciousness by anesthetics can lead to loss of spines and eventual loss of neurons.

Explained based on the semblance hypothesis (Vadakkan, 2015a). Conversion of IPLs to inter-neuronal interspine fusion leading to degeneration is a testable mechanism.


More years of education (increased number of associative learning events) reduces dementia risk (Maccora et al., 2020).

Should be able to explain whether redundant learning-induced changes get induced by prolonged learning events.

Explained based on the semblance hypothesis (Vadakkan, 2013; 2019a). Redundant IPLs form during different learning events as new neurons get inserted into the circuit.


Specific brain regions appear to be associated with specific functions based on the lesions/ lesion studies.

These are most likely locations of converging fiber tracts or converging locations of specific input signals responsible for those functions.

It was possible to induce long-term potentiation (LTP) of different strengths from different locations of convergence inputs. Hippocampus having convergence of all the sensory inputs has shown maximum strength of LTP.


Astrocytic pedocytes cover less than 50% of peri-synaptic area in nearly 60% of the synapses in the CA1 region of hippocampus (Ventura and Harris, 1999).

Hippocampal mechanism of learning & memory must explain the suitability of distribution of astrocytic processes.

Explained based on the semblance hypothesis (Vadakkan, 2019a). Remaining free area of the spines favor inter-neuronal inter-spine interactions that form IPLs.


Present nervous systems have evolved over millions of years and are also the results of certain accidental coincidences.

It is expected to become possible to explain how the circuitry that provides all the features can be evolved through simple steps of variations and selection.

Explained based on the semblance hypothesis (Vadakkan, 2020). Sparking of the first-person inner sensations of all the features of an item in the environment on arrival of the fastest or first sensory stimulus from that item started providing survival advantage to animals in a predator-prey environment.


Dye diffuses from one neuronal cell to another as the cortical neurons move from periventricular region towards their destination indicating formation of an inter-cellular fusion pore (Bittman et al., 1997). This is followed by death of nearly 70% of these cells and survival of the remaining 30% cells.

It is expected to become possible to explain how an event of inter-cellular fusion leads to selection of variants that prevent further inter-cellular fusion. Since neurons cannot divide further, a transient stage of fusion is expected to trigger fusion prevention mechanism in the surviving neuronal cells. It is also necessary to explain whether this mechanism has any role in the unique functional property of generation of first-person inner sensations in the nervous system.

Explained based on the semblance hypothesis (Vadakkan, 2020). Dye diffusion indicates formation of fusion pores between neuronal cells. The first occurrence of inter-neuronal fusion is likely due to changes in membrane composition or lack of checkpoint mechanisms to arrest hemifusion from progressing to fusion.


Following the above stage where dye diffusion is observed, significant neuronal death (70%) (Blaschke et al., 1996) and spine loss (13 to 20%) are observed.

There is a high probability that the surviving cells have acquires an adaptation.

Explained based on the semblance hypothesis (Vadakkan, 2020). Following death of 70% cells, an adaptation occurring in the surviving cells most likely prevents any future coupling between neurons that may result in inter-neuronal fusion. This adaptation is suitable for maintaining IPLs for generating useful functions.


Higher brain functions take place in a narrow range of frequency of oscillating extracellular potentials as evidenced by EEG (Rusalova, 2006).

a) Both the mechanism for learning and memory retrieval contributes vector components of the oscillating extracellular potentials. b) The specific mechanism for both learning and memory retrieval depends on the frequency of oscillating extracellular potentials.

Explained based on the semblance hypothesis (Vadakkan, 2010a; 2013).


Artificial triggering of spikes in one neuron in the cortex causes spikes in a group of neighboring neurons in the same neuronal order located at short distance (25–70µm) from the stimulated neuron (Chettih & Harvey, 2019).

It should be possible to explain a mechanism that can lead to lateral spread of firing between neurons of the same neuronal order within a short radius. Need an explanation for a mechanism through a path other than trans-synaptic route.

One explanation is propagation of depolarization across the IPLs between spines that belong to different neurons (Vadakkan, 2013). This also explains why only  sparsely located neurons get fired, correlated in time.


The protein complexin blocks SNARE-mediated fusion by arresting the intermediate stage of hemifusion. Complexin is present in the spines. But docked vesicles are not found inside the spines (in contrast to what is observed in the presynaptic terminal). This leaves the question, "Which inter-membrane fusion is getting arrested by complexin?"

It is necessary to explain an inter-membrane fusion process that can be mediated by SNARE proteins and blocked by complexin by arresting the process at or before the intermediate stage of hemifusion in the spines.

SNARE proteins provide energy for bringing together membranes against repulsive charges and overcoming energy barrier between abutted membranes (Oelkers et al., 2016). They also generate force to pull together abutted membranes as tightly as possible (Hernandez et al., 2012). By initiating the fusion process by supplying energy (Jahn and Scheller, 2006), SNARE proteins can lead to the formation of characteristic hemifusion intermediates (Lu et al., 2005; Giraudo et al., 2005; Liu et al., 2008). Protein complexin present within the postsynaptic terminals (Ahmad et al., 2012) is known to interact with the neuronal SNARE core complex to arrest fusion at the stage of hemifusion (Schaub et al., 2006).


Transcriptomic analyses show heterogeneity of even adjacent neurons of the same type in the cortex (Kamme et al., 2003).

This indicates that any mixing of the contents between these neurons is fatal to them. Hence, there will be a robust mechanism to prevent intercellular fusion.

Different mRNA profiles of adjacent neurons of even the same type indicate that any cytoplasmic content mixing will lead to homeostatic mechanisms such as spine or neuronal loss to prevent it (Vadakkan, 2016c). Ultimate purpose of it is to restrict structural aspect of IPLs to inter-membrane hemifusion.


Heterogeneity in clinical features and pathological changes in Alzheimer's disease (& other neurodegenerative disorders).

1. Many factors are likely involved in the operational mechanism. 2. There will be a universal mechanism that involves different neuronal types. Failure of any of these can explain heterogeneity.

A common mechanism is pathological conversion of normal maximum limit of hemifusion to pathological fusion. Clinical features depend on a) locations of IPL fusion that can damage spines & neurons, and b) formation of non-specific IPLs at different locations (Vadakkan, 2016c).


In excitatory neurons, spine depolarization can occur even without dendritic depolarization (Beaulieu-Laroche et al., 2018a; Beaulieu-Laroche et al., 2018b).

Why did such a mechanism get selected? What is the functional significance of depolarization of the spine head? Is there any link between depolarization of the spine heads, oscillating extracellular potentials & different brain functions?

IPL mechanism that generates units of inner sensation needs only depolarization of spines. Lack of firing of the postsynaptic neuron will lead to lack of motor output while units of inner sensation occur at an inter-LINKed spine (Vadakkan, 2013; 2019a).


The histological features of amyloid (senile) plaques and neurofibrillary tangles observed in normal aging (Anderton, 1997) are also the pathological features in Alzheimer's disease & several other disorders in the spectrum of neurodegenerative disorders.

A mechanistic explanation for how & why intracellular neurofibrillary tangles & extracellular plaques that are key pathological features in neurodegenerative disorders are observed in normal aging (but without symptoms).

The last stage of IPL formation is hemifusion, which is an intermediate stage of fusion. Various factors such as viral fusion proteins and membrane compositional changes can overcome the check point mechanisms converging IPLs to inter-neuronal inter-spine fusion. This will cause cytoplasmic content mixing. Since expression profiles of even adjacent neurons of same type are different, it leads to neuronal damage  (Vadakkan 2016c).


Therapeutic agents developed for treating seemingly unrelated neurological diseases such as seizure disorders, Parkinson's disease, spasticity, and hallucinations can alleviate different headache pains.

Explanations for mechanisms of different disorders & the operational mechanism of the system should provide interconnected explanations for the effectiveness of therapeutic agents in different headaches.

By inhibiting voltage-gated sodium channels, it reduces neuronal excitability & prevent rapid IPL formation preventing seizures, prevents IPL formation between spines of spiny neurons of basal ganglia, reduce inputs via IPLs to upper motor neurons reducing spasticity, reverse/inhibit IPLs inhibiting/reducing inner sensation of headache pains.


Since learning is expected to generate certain new circuit connections, the circuit elements (like on a printed circuit board) must remain separate from each other.

Properties of both neuronal membranes and extracellular matrix should match with the new circuit connections, functional properties imparted by them and their reversal.

Even though extracellular matrix space seems negligible between the membranes, hydration layer between the lipid membranes shows high energy barrier in artificial systems (Rand and Parsegian, 1984; Martens and McMahon, 2008; Harrison, 2015).


"Representational drift" - meaning that when a brain function is repeated, set of neurons that fire changes with time (Schoonover et al., 2021; Marks& Goard, 2021; Deitch et al., 2021).

In the case of memory, it is necessary to show redundancy in its operational mechanism, presence of a common integration mechanism and shift in the locations from where function occurs.

Correlation between a brain function and neuronal firing will be true for those neurons that are being held at sub-threshold activation state and receive additional potentials through the same IPLs. Subthreshold activation state of a neuron can be affected by several factors. Additional learning events can lead to the formation of new IPLs. These can change the set of neurons that fire (Vadakkan, 2019a).


The controversial views (pdf) expressed by Camillo Golgi against Ramón y Cajal's interpretations of results obtained from modified Golgi staining protocols.

The chemistry behind the modification of original Golgi staining protocol must be able to provide reasons for this controversy. Such an explanation is expected to become possible when we understand the operational mechanism of the brain.

Spines within the islets of inter-LINKed spines are connected via an oxidation state dependent manner (Vadakkan, 2022).


Formation of new neurons in the hippocampus, especially in non-stationary environments.

The operational mechanism should be able to explain functional advantage provided by insertion of new neurons.

Both input and output connections of new neurons will continuously alter the existing circuitry. Repetition of same associative learning will make new IPLs at higher neuronal orders increasing number of sparse storage mechanisms (Vadakkan, 2011a).


Loss of spines and formation of new spines during learning (Frank et al., 2018).


There is a mechanism that leads to loss of spines during learning. Formation of new spines should accomplish something new that can facilitate further learning.

The last stage of permitted intermembrane interaction leading to IPL formation is inter-membrane hemifusion, which is an intermediate stage of membrane fusion. Several factors can overcome the checkpoint needed to arrest the changes at the stage of hemifusion.


Generalization is seen in the Transformers of the large language models (LLMs) that use neural networks organized using a hidden layer.

A mechanistic explanation is needed to explain where and how the hidden layer operates. This must explain how the system generates an appropriate output in response to a new prompt that the system never got exposed in the past.

Basic concept of “islet of inter-LINKed spines” in the semblance hypothesis matches with that of the “attention heads” in the Transformers of LLMs (Vadakkan, 2024).

99 Gamma waveforms of oscillating extracellular potentials in the layers 2 & 3, & slower alpha and beta waves in deeper layers of the cortex (Mendoza-Halliday et al., 2024). Reasons are necessary to provide explanation for high frequency of oscillations in layers 2 &3 & low frequency in deeper layers Apical tuft region of neurons of all the cortical layers are attached to the marginal zone close to the pial layer (innermost covering layer of the cortex). Layer 1 cortical neurons that are mostly GABAergic send horizontal processes interconnecting several postsynaptic terminals of apical tufts. Hence, dendritic arbors of neurons of all the cortical layers overlap maximally in layer 2, followed by layers 3, 4, 5, and 6 in decreasing order. This increases the probability of formation of maximum number of synapses (vertical component) & inter-neuronal inter-spine interactions (horizontal component) in layer 2. This in turn explains high frequency of oscillations of potentials across neurons that are connected vertically (through synapses) & horizontally (through IPLs) in a decreasing order from layer 2 to 6 (Vadakkan, 2015).


IPL: Inter-postsynaptic functional LINK

Foot note1 If we provide a set of colors and picture against each color and ask people to learn the association in one second, most people will be able to  learn two or more associations during this period. This means that associative learning can take place in milliseconds. What type of a change can occur in less than one second?


Foot note2 Around 30 specific experimental findings are related to the association between LTP, ability to learn, behavioral motor action at the of memory retrieval and biochemical changes following learning. It was possible to satisfy all the constraints arising from those findings in an interconnected manner.

Table 2. Features of the system from different levels that need to be explained independently and by an inter-connectable manner using a derived solution. In other words, these constraints permit us to ask the question, "What should be the foundational operation that can satisfy all these constraints?" Even though several possibilities can be excluded (for example, biochemical reactions that occur slower than the physiological timescales of milliseconds during which learning takes place (using which memory needs to be retrieved) that can help exclude candidacy of several biochemical intermediates such as storage molecules), a systematic approach is necessary to find the correct solution. Please note that the listed findings are so disparate, and the constraints offered by them are so strong that there can only be one unique solution. In other words, this unique solution for the system should be compatible with all the previous experimental observations. Constraints provided by each of the observations help to narrow down the possibilities to arrive at the solution. A subset of the above list of observations can be used to derive the solution and the rest of the features can be used to verify whether the derived solution is correct or not. Please note that we cannot arrive at a solution using a few mathematical equations. Once we have a unitary solution, we need to search for the principle of their integration where mathematics is expected to have a role.


A complete understanding of the operational mechanism leading to the first-person properties will only be achieved by carrying out the gold standard test of its replication in engineered systems. Even though replication of motor activities to produce behaviorally equivalent machines may seem adequate, the work will not be complete until first-person properties of the mind are understood. Engineering challenges with this approach include devising methods to convert the first-person accessible internal sensations to appropriate readouts. Experiments to translate theoretically feasible mechanisms of its formation both by computational and engineering methods are required. Feasibility to explain various brain functions both from first-person and third-person perspectives qualifies it as a testable hypothesis. Research findings from different laboratories have been examined in terms of the semblance hypothesis. The present work resulted from curiosity to understand the order behind the seemingly complex brain functions. In this attempt, I have used some freedom to seek a new basic principle in order to put the pieces of the puzzle together. This work wouldn't have become possible without a large amount of research work painstakingly carried out by many researchers over many years. Even though the present hypothesis is compatible with experimental data, it must be considered unproven until further verification of its testable predictions are made.


Video presentations

1. A testable hypothesis of brain functions

2. How to study inner sensations? Examples from mathematics

3. Neurons and Synapses

4. List of third-person findings and the derivation of the solution for the nervous system

5. Constraints to work with

6. Induction of units of inner sensation

7. Why do we need to sleep?

8. Potentential mechanism for neurodegeneration

9. LTP: An explanation by semblance hypothesis

10. A framework for consciousness

11. A potential mechanism of anaesthetic agents


The challenge: "What I cannot create (replicate), I do not understand" Richard Feynman. The rigor with which we should try to solve the nervous system must be with an intention to replicate its mechanism in an engineered system. Everything else will follow.

The reality: We are being challenged to find a scientific method to study the unique function of the nervous system - how different inner sensations are being generated in the brain concurrent with different third-person observed findings. We cannot directly study them using biological systems. But we can use all the observations to try to solve the system theoretically, followed by verifying its predictions.  

The optimism:“What are the real conditions that the solution must satisfy?” If we can get that right, then we can try and figure out what the solution is" – Murray Gell–Mann

The expectation: We are likely able to solve the mechanism of the nervous system functions in multiple steps. First, using constraints offered by all the observations, it is necessary to derive a solution (most likely a first principle) that can unify those observations. This can be followed by further verifications by triangulation methods and examining comparable circuitries in different animal species. Once identified and verified, we can expect to replicate the mechanism in engineered systems.

The advice: "Nothing in life is to be feared, it is only to be understood. Now is the time to understand more, so that we may fear less." Marie Curie

The hope: We will give everything we can. Together we will explore it!