Tag Archives: Artificial Intelligence

The predictive brain

While significant progress has been made in the field of neuroscience and in particular in the neural circuits that support perception and motor activity, the understanding of neural structures, how they encode information, and establish the mechanisms of learning is still under investigation.

Digital audio and image processing techniques and advances in artificial intelligence (AI) are a source of inspiration for understanding these mechanisms. However, it seems clear that these ideas are not directly applicable to brain functionality.

Thus, for example, the processing of an image is static, since digital sensors provide complete images of the scene. In contrast, the information encoded by the retina is not homogeneous, with large differences in resolution between the fovea and the surrounding areas, so that the image composition is necessarily spatially segmented.

But these differences are much more pronounced if we consider that this information is dynamic in time. In the case of digital video processing, it is possible to establish a correlation of the images that make up a sequence. A correlation that in the case of the visual system is much more complex, due to the spatial segmentation of the images and how this information is obtained using the saccadic movements of the eyes.

The information generated by the retina is processed by the primary visual cortex (V1) which has a well-defined map of spatial information and also performs simple feature recognition functions. This information progresses to the secondary visual cortex (V2) which is responsible for composing the spatial information generated by saccadic eye movement.

This structure has been the dominant theoretical framework, in what has been termed the hierarchical feedforward model [1]. However, certain neurons in V1 and V2 regions have been found to have a surprising response. They seem to know what is going to happen in the immediate future, activating as if they could perceive new visual information without it having been produced by the retina [2], in what is defined as Predictive Processing (PP)  [3], and which is gaining influence in cognitive neuroscience, although it is criticized for lacking empirical support to justify it.

For this reason, the aim of this post is to analyze this behavior from the point of view of signal processing techniques and control systems, which show that the nervous system would not be able to interact with the surrounding reality unless it performs PP functions.

A brief review of control systems

The design of a control system is based on a mature technique [4], although the advances in digital signal processing produced in the last decades allow the implementation of highly sophisticated systems. We will not go into details about these techniques and will only focus on the aspects necessary to justify the possible PP performed by the brain.

Thus, a closed-loop control system is composed of three fundamental blocks:

  • Feedback: This block determines the state of the target under control.
  • Control: Determines the actions to be taken based on the reference and the information on the state of the target.
  • Process: Translates the actions determined by the control to the physical world of the target.

The functionality of a control system is shown in the example shown in the figure. In this case the reference is the position of the ball and the target is for the robot to hit the ball accurately.

The robot sensors must determine in real-time the relative position of the ball and all the parameters that define the robot structure (feedback). From these, the control must determine the robot motion parameters necessary to reach the target, generating the control commands that activate the robot’s servomechanisms.

The theoretical analysis of this functional structure allows determining the stability of the system, which establishes its capacity to correctly develop the functionality for which it has been designed. This analysis shows that the system can exhibit two extreme cases of behavior. To simplify the reasoning, we will eliminate the ball and assume that the objective is to reach a certain position.

In the first case, we will assume that the robot has a motion capability such that it can perform fast movements without limitation, but that the measurement mechanisms that determine the robot’s position require a certain processing time Δt. As a consequence, the decisions of the control block are not in real-time since the decisions at t = ti actually correspond to t = ti-Δt, where Δt is the time required to process the information coming from the sensing mechanisms. Therefore, when the robot approaches the reference point the control will make decisions as if it were somewhat distant, which will cause the robot to overshoot the position of the target. When this happens, the control should correct the motion by turning back the robot’s trajectory. This behavior is defined as an underdamped regime.

Conversely, if we assume that the measurement system has a fast response time, such that Δt≊0, but that the robot’s motion capability is limited, then the control will make decisions in real-time, but the approach to the target will be slow until the target is accurately reached. Such behavior is defined as an overdamped regime.

At the boundary of these two behaviors is the critically damped regime that optimizes the speed and accuracy to reach the target. The behavior of these regimes is shown in the figure.

Formally, the above analysis corresponds to systems in which the functional blocks are linear. The development of digital processing techniques allows the implementation of functional blocks with a nonlinear response, resulting in much more efficient control systems in terms of response speed and accuracy. In addition, they allow the implementation of predictive processing techniques using the laws of mechanics. Thus, if the reference is a passive entity, its trajectory is known from the initial conditions. If it is an active entity, i.e. it has internal mechanisms that can modify its dynamics, heuristic functions, and AI can be used  [5].

The brain as a control system

As the figure below shows, the ensemble formed by the brain, the motor organs, and the sensory organs comprises a control system. Consequently, this system can be analyzed with the techniques of feedback control systems.

For this purpose, it is necessary to analyze the response times of each of the functional blocks. In this regard, it should be noted that the nervous system has a relatively slow temporal behavior [6]. Thus, for example, the response time to initiate movement in a 100-meter sprint is 120-165 ms. This time is distributed in recognizing the start signal, the processing time of the brain to interpret this signal and generate the control commands to the motor organs, and the start-up of these organs. In the case of eye movements toward a new target, the response time is 50-200 ms. These times give an idea of the processing speed of the different organs involved in the different scenarios of interaction with reality.

Now, let’s assume several scenarios of interaction with the environment:

  • A soccer player intending to hit a ball moving at a speed of 10 km/hour. In a time of 0.1 s. the ball will have moved 30 cm. 
  • A tennis player who must hit a ball moving at 50 km/hour. In a time of 0.1 s. the ball will have displaced 150 cm. 
  • Grip a motionless cup by moving the hand at a speed of 0.5 m/s. In a time of 0.1 s. the hand will have moved 5 cm.

These examples show that if the brain is considered as a classical control system, it is practically impossible to obtain the necessary precision to justify the behavior of the system. Thus, in the case of the soccer player, the information obtained by the brain from the sensory organs, in this case, the sight, will be delayed in time, providing a relative position of the foot concerning the ball with an error of the order of centimeters, so that the ball strike will be very inaccurate.

The same reasoning can be made in the case of the other two proposed scenarios, so it is necessary to investigate the mechanisms used by the brain to obtain an accuracy that justifies its actual behavior, much more accurate than that provided by a control system based on the temporal response of neurons and nerve tissue.

To this end, let’s assume the case of grasping the cup, and let’s do a simple exercise of introspection. If we close our eyes for a moment we can observe that we have a precise knowledge of the environment. This knowledge is updated as we interact with the environment and the hand approaches the cup. This spatiotemporal behavior allows predicting with the necessary precision what will be the position of the hand and the cup at any moment, despite the delay produced by the nervous system.

To this must be added the knowledge acquired by the brain about space-time reality and the laws of mechanics. In this way, the brain can predict the most probable trajectory of the ball in the tennis player’s scenario. This is evident in the importance of training in sports activities since this knowledge must be refreshed frequently to provide the necessary accuracy. Without the above prediction mechanisms, the tennis player would not be able to hit the ball.

Consequently, from the analysis of the behavior of the system formed by the sensory organs, the brain, and the motor organs, it follows that the brain must perform PP functions. Otherwise, and as a consequence of the response time of the nervous tissue, the system would not be able to interact with the environment with the precision and speed shown in practice. In fact, to compensate for the delay introduced by the sensory organs and their subsequent interpretation by the brain, the brain must predict and advance the commands to the motor organs in a time interval that can be estimated at several tens of milliseconds.

The neurological foundations of prediction

As justified in the previous section, from the temporal response of the nervous tissue and the behavior of the system formed by the sensory organs, the brain, and the motor organs, it follows that the brain must support two fundamental functions: encoding and processing reference frames of the surrounding reality and performing Predictive Processing.

But what evidence is for this behavior? It has been known for several decades that there are neurons in the entorhinal cortex and hippocampus that respond to a spatial model, called grid cells [7]. But recently it has been shown that in the neocortex there are structures capable of representing reference frames and that these structures can render both a spatial map and any other functional structure needed to represent concepts, language, and structured reasoning [8].

Therefore, the question to be resolved is how the nervous system performs PP. As already advanced, PP is a disputed functionality because of its lack of evidence. The problem it poses is that the number of neurons that exhibit predictive behavior is very small compared to the number of neurons that are activated as a consequence of a stimulus.

The answer to this problem may lie in the model proposed by Jeff Hawkins and Subutai Ahmad [9] based on the functionality of pyramidal neurons [10], whose function is related to motor control and cognition, areas in which PP should be fundamental.

The figure below shows the structure of a pyramidal neuron, which is the most common type of neuron in the neocortex. The dendrites close to the cell body are called proximal synapses so that the neuron is activated if they receive sufficient excitation. The nerve impulse generated by the activation of the neuron propagates to other neurons through the axon, which is represented by an arrow.

This description corresponds to a classical view of the neuron, but pyramidal neurons have a much more complex structure. The dendrites radiating from the central zone are endowed with hundreds or thousands of synapses, called distal synapses so approximately 90% of the synapses are located on these dendrites. Also, the upper part of the figure shows dendrites that have a longer reach, which have feedback functionality.

The remarkable thing about this type of neuron is that if a group of synapses of a distal dendrite close to each other receives a signal at the same time, a new type of nerve impulse is produced that propagates along the dendrite until it reaches the body of the cell. This causes an increase in the voltage of the cell, but without producing its activation, so it does not generate a nerve impulse towards the axon. The neuron remains in this state for a short period, returning to its relaxed state.

The question is: What is the purpose of these nerve impulses from the dendrites if they are not powerful enough to produce cell activation? This has been an unknown that is intended to be solved by the model proposed by Hawkins and Ahmad [9], which proposes that the nerve impulses in the distal dendrites are predictions.

This means that a dendritic impulse is produced when a set of synapses close to each other on a distal dendrite receive inputs at the same time, and it means that the neuron has recognized a pattern of activity determined by a set of neurons. When the pattern of activity is detected, a dendritic impulse is created, which raises the voltage in the cell body, putting the cell into what we call a predictive state.

The neuron is then ready to fire. If a neuron in the predictive state subsequently receives sufficient proximal input to create an action potential to fire it, then the neuron fires slightly earlier than it would if the neuron were not in the predictive state.

Thus, the prediction mechanism is based on the idea that multiple neurons in a minicolumn [11] participate in the prediction of a pattern, all of them entering a prediction state, such that when one of them fires it inhibits the firing of the rest. This means that in a minicolumn hundreds or thousands of predictions are made simultaneously over a certain control scenario, such that one of the predictions will prevail over the rest, optimizing the accuracy of the process. This justifies the fact of the small number of predictive events observed versus the overall neuronal activity and also explains why unexpected events or patterns produce greater activity than more predictable or expected events.

If the neural structure of the minicolumns is taken into account, it is easy to understand how this mechanism involves a large number of predictions for the processing of a single pattern, and it can be said that the brain is continuously making predictions about the environment, which allows real-time interaction.

The PP from the point of view of AI

According to the above analysis, it can be concluded that the PP performed by the brain within a time window, of the order of tens of milliseconds, is fundamental for the interaction with the surrounding reality, synchronizing this reality with the perceived reality. But this ability to anticipate perceived events requires other mechanisms such as the need to establish reference frames as well as the ability to recognize patterns.

In the subject raised, it is evident the need to have reference frames in which objects can be represented, such as the dynamic position of the motor organs and of the objects with which to interact. In addition to this, the brain must be able to recognize such objects.

But these capabilities are common to all types of scenarios, although it is perhaps more appropriate to use the term model as an alternative to a reference frame since it is a more general concept. Thus, for example, in verbal communication, it is necessary to have a model that represents the structure of language, as well as an ability to recognize the patterns encoded in the stimuli perceived through the auditory system. In this case, the PP must play a fundamental role, since prediction allows for greater fluency in verbal communication, as is evident when there are delays in a communication channel. This is perhaps most evident in the synchronism necessary in musical coordination.

The enormous complexity of the nervous tissue and the difficulty to empirically identify these mechanisms can be an obstacle to understanding their behavior. For this reason, AI is a source of inspiration [12] since, using different neural network architectures, it shows how models of reality can be established and predictions can be made about this reality.

It should be noted that these models do not claim to provide realistic biological models. Nevertheless, they are fundamental mathematical models in the paradigm of machine learning and artificial intelligence and are a fundamental tool in neurological research. In this sense, it is important to highlight that PP is not only a necessary functionality for the temporal prediction of events, but as shown by artificial neural networks pattern recognition is intrinsically a predictive function.

This may go unnoticed in the case of the brain since pattern recognition achieves such accuracy that it makes the concept of prediction very diluted and appears to be free of probabilistic factors. In contrast, in the case of AI, mathematical models make it clear that pattern recognition is probabilistic in nature and practical results show a diversity of outcomes.

This diversity depends on several factors. Perhaps the most important is its state of development, which can still be considered very primitive, compared to the structural complexity, processing capacity, and energy efficiency of the brain. This means that AI applications are oriented to specific cases where it has shown its effectiveness, such as in health sciences [13] or in the determination of protein structures [14].

But without going into a deeper analysis of these factors, what can be concluded is that the functionality of the brain is based on the establishment of models of reality and the prediction of patterns, one of its functions being temporal prediction, which is the foundation of PP. 


[1]J. DiCarlo, D. Zoccolan y N. Rust, «How does the brain solve visual object recognition?,» Neuron, vol. 73, pp. 415-434, 2012.
[2]A. Clark, «Whatever next? Predictive brains, situated agents, and the future of cognitive science,» Behav. Brain Sci., vol. 34, p. 181–204, 2013.
[3]W. Wiese y T. Metzinger, «Vanilla PP for philosophers: a primer on predictive processing.,» In Philosophy and Predictive Processing. T. Metzinger &W.Wiese, Eds., pp. 1-18, 2017.
[4]G. F. Franklin, J. D. Powell y A. Emami-Naeini, Feedback Control of Dynamic Systems, Pearson; 8a edición, 2019.
[5]C. Su, S. Rakheja y H. Liu, «Intelligent Robotics and Applications,» de 5th International Conference, ICIRA, Proceedings, Part II, Montreal, QC, Canada, 2012.
[6]A. Roberts, R. Borisyuk, E. Buhl, A. Ferrario, S. Koutsikou, W.-C. Li y S. Soffe, «The decision to move: response times, neuronal circuits and sensory memory in a simple vertebrate,» Proc. R. Soc. B, vol. 286: 20190297, 2019.
[7]M. B. Moser, «Grid Cells, Place Cells and Memory,» de Nobel Lecture. Aula Medica, Karolinska Institutet, Stockholm, http://www.nobelprize.org/prizes/medicine/2014/may-britt-moser/lecture/, 2014.
[8]M. Lewis, S. Purdy, S. Ahmad y J. Hawkings, «Locations in the Neocortex: A Theory of Sensorimotor Object Recognition Using Cortical Grid Cells,» Frontiers in Neural Circuits, vol. 13, nº 22, 2019.
[9]J. Hawkins y S. Ahmad, «Why Neurons Have Tousands of Synapses, Theory of Sequence Memory in Neocortex,» Frontiers in Neural Circuits, vol. 10, nº 23, 2016.
[10]G. N. Elston, «Cortex, Cognition and the Cell: New Insights into the Pyramidal Neuron and Prefrontal Function,» Cerebral Cortex, vol. 13, nº 11, p. 1124–1138, 2003.
[11]V. B. Mountcastle, «The columnar organization of the neocortex,» Brain, vol. 120, p. 701–722, 1997.
[12]F. Emmert-Streib, Z. Yang, S. Tripathi y M. Dehmer, «An Introductory Review of Deep Learning for Prediction Models With Big Data,» Front. Artif. Intell., 2020.
[13]A. Bohr y K. Memarzadeh, Artificial Intelligence in Healthcare, Academic Press, 2020.
[14]E. Callaway, «‘It will change everything’: DeepMind’s AI makes gigantic leap in solving protein structures,» Nature, nº 588, pp. 203-204, 2020.

Consciousness from the point of view of AI

The self-awareness of human beings, which constitutes the concept of consciousness, has been and continues to be an enigma faced by philosophers, anthropologists and neuroscientists. But perhaps most suggestive is the fact that consciousness is a central concept in human behavior and that being aware of it does not find an explanation for it.

Without going into details, until the modern age the concept of consciousness had deep roots in the concept of soul and religious beliefs, often attributing to divine intervention in the differentiation of human nature from other species.

The modern age saw a substantial change based on Descartes’ concept “cogito ergo sum ( I think, therefore I am”) and later on the model proposed by Kant, which is structured around what are known as “transcendental arguments” [1].

Subsequently, a variety of schools of thought have developed, among which dualistic, monistic, materialistic and neurocognitive theories stand out. In general terms, these theories focus on the psychological and phenomenological aspects that describe conscious reality. In the case of neurocognitive theories, neurological evidence is a fundamental pillar. But ultimately, all these theories are abstract in nature and, for the time being, have failed to provide a formal justification of consciousness and how a “being” can develop conscious behavior, as well as concepts such as morality or ethics.

One aspect that these models deal with and that brings into question the concept of the “cogito” is the change of behavior produced by brain damage and that in some cases can be re-educated, which shows that the brain and the learning processes play a fundamental role in consciousness.

In this regard, advances in Artificial Intelligence (AI) [2] highlight the formal foundations of learning, by which an algorithm can acquire knowledge and in which neural networks are now a fundamental component. For this reason, the use of this new knowledge can shed light on the nature of consciousness.

The Turing Test paradigm

To analyze what may be the mechanisms that support consciousness we can start with the Turing Test [3], in which a machine is tested to see if it shows a behavior similar to that of a human being.

Without going into the definition of the Turing Test, we can assimilate this concept to that of a chatbot, as shown in Figure 1, which can give us an intuitive idea of this concept. But we can go even further if we consider its implementation. This requires the availability of a huge amount of dialogues between humans, which allows us to train the model using Deep Learning techniques [4]. And although it may seem strange, the availability of dialogues is the most laborious part of the process.

Figure 1. Schematic of the Turing Test

Once the chatbot has been trained, we can ask about its behavior from a psychophysical point of view. The answer seems quite obvious, since although it can show a very complex behavior, this will always be a reflex behavior, even though the interlocutor can deduce that the chatbot has feelings and even an intelligent behavior. The latter is a controversial issue because of the difficulty of defining what constitutes intelligent behavior, which is highlighted by the questions: Intelligent? Compared to what?

But the Turing Test only aims to determine the ability of a machine to show human-like behavior, without going into the analysis of the mechanisms to establish this functionality.

In the case of humans, these mechanisms can be classified into two sections: genetic learning and neural learning.

Genetic learning

Genetic learning is based on the learning capacity of biology to establish functions adapted to the processing of the surrounding reality. Expressed in this way it does not seem an obvious or convincing argument, but DNA computing [5] is a formal demonstration of the capability of biological learning. The evolution of capabilities acquired through this process is based on trial and error, which is inherent to learning. Thus, biological evolution is a slow process, as nature shows.

Instinctive reactions are based on genetic learning, so that all species of living beings are endowed with certain faculties without the need for significant subsequent training. Examples are the survival instinct, the reproductive instinct, and the maternal and paternal instinct. These functions are located in the inner layers of the brain, which humans share with vertebrates.

We will not go into details related to neuroscience [6], since the only thing that interests us in this analysis is to highlight two fundamental aspects: the functional specialization and plasticity of each of its neural structures. Thus, structure, plasticity and specialization are determined by genetic factors, so that the inner layers, such as the limbic system, have a very specialized functionality and require little training to be functional. In contrast, the external structures, located in the neocortex, are very plastic and their functionality is strongly influenced by learning and experience.

Thus, genetic learning is responsible for structure, plasticity and specialization, whereas neural learning is intimately linked to the plastic functionality of neural tissue.

A clear example of functional specialization based on genetic learning is the space-time processing that we share with the rest of higher living beings and that is located in the limbic system. This endows the brain with structures dedicated to the establishment of a spatial map and the processing of temporal delay, which provides the ability to establish trajectories in advance, vital for survival and for interacting with spatio-temporal reality.

This functionality has a high degree of automaticity, which makes its functional capacity effective from the moment of birth. However, this is not exactly the case in humans, since these neural systems function in coordination with the neocortex, which requires a high degree of neural training.

Thus, for example, this functional specialization precludes visualizing and intuitively understanding geometries of more than three spatial dimensions, something that humans can only deal with abstractly at a higher level by means of the neocortex, which has a plastic functionality and is the main support for neural learning.

It is interesting to consider that the functionality of the neocortex, whose response time is longer than that of the lower layers, can interfere in the reaction of automatic functions. This is clearly evident in the loss of concentration in activities that require a high degree of automatism, as occurs in certain sports activities. This means that in addition to having an appropriate physical capacity and a well-developed and trained automatic processing capacity, elite athletes require specific psychological preparation.

This applies to all sensory systems, such as vision, hearing, balance, in which genetic learning determines and conditions the interpretation of information coming from the sensory organs. But as this information ascends to the higher layers of the brain, the processing and interpretation of the information is determined by neural learning.

This is what differentiates humans from the rest of the species, being endowed with a highly developed neocortex, which provides a very significant neural learning capacity, from which the conscious being seems to emerge.

Nevertheless, there is solid evidence of the ability to feel and to have a certain level of consciousness in some species. This is what has triggered a movement for legal recognition of feelings in certain species of animals, and even recognition of personal status for some species of hominids.

Neural learning: AI as a source of intuition

Currently, AI is made up of a set of mathematical strategies that are grouped under different names depending on their characteristics. Thus, Machine Learning (ML) is made up of classical mathematical algorithms, such as statistical algorithms, decision trees, clustering, support vector machine, etc. Deep Learning, on the other hand, is inspired by the functioning of neural tissue, and exhibits complex behavior that approximates certain capabilities of humans.

In the current state of development of this discipline, designs are reduced to the implementation and training of specific tasks, such as automatic diagnostic systems, assistants, chatbots, games, etc., so these systems are grouped in what is called Artificial Narrow Intelligence.

The perspective offered by this new knowledge makes it possible to establish three major categories within AI:

  • Artificial Narrow Intelligence.
  • Artificial General Intelligence. AI systems with a capacity similar to that of human beings.
  • Artificial Super Intelligence: Self-aware AI systems with a capacity equal to or greater than that of human beings. 

The implementation of neural networks used in Deep Learning is inspired by the functionality of neurons and neural tissue, as shown in Figure 2 [7]. As a consequence, the nerve stimuli coming from the axon terminals that connect to the dendrites (synapses) are weighted and processed according to the functional configuration of the neuron acquired by learning, producing a nerve stimulus that propagates to other neurons, through the terminal axons.

Figure 2. Structure of a neuron and mathematical model

Artificial neural networks are structured by creating layers of the mathematical neuron model, as shown in Figure 3. A fundamental issue in this model is to determine the mechanisms necessary to establish the weighting parameters Wi in each of the units that form the neural network. Neural mechanisms could be used for this purpose. However, although there is a very general idea of how the functionality of the synapses is configured, the establishment of the functionality at the neural network level is still a mystery.

Figure 3. Artificial Neural Network Architecture

In the case of artificial neural networks, mathematics has found a solution that makes it possible to establish the Wi values, by means of what is known as supervised learning. This requires having a dataset in which each of its elements represents a stimulus X i and the response to this stimulus Y i. Thus, once the Wi values have been randomly initialized, the training phase proceeds, presenting each of the X i stimuli and comparing the response with the Y i values. The errors produced are propagated backwards by means of an algorithm known as backpropagation.

Through the sequential application of the elements of a training set belonging to the dataset in several sessions, a state of convergence is reached, in which the neural network achieves an appropriate degree of accuracy, verified by means of a validation set of elements belonging to the dataset that are not used for training.

An example is much more intuitive to understand the nature of the elements of a dataset. Thus, in a dataset used in the training of autonomous driving systems, X i correspond to images in which patterns of different types of vehicles, pedestrians, public roads, etc. appear. Each of these images has a category Y i associated with it, which specifies the patterns that appear in that image. It should be noted that in the current state of development of AI systems, the dataset is made by humans, so learning is supervised and requires significant resources.

In unsupervised learning the category Y i is generated automatically, although its state of development is very incipient. A very illustrative example is the Alpha Zero program developed by DeepMind [8], in such a way that learning is performed by providing it with the rules of the game (chess, go, shogi) and developing against itself matches, in such a way that the moves and the result configure (X i , Y i). The neural network is continuously updated with these results, sequentially improving its behavior and therefore the new results (X i , Y i), reaching a superhuman level of play.

It is important to note that in the case of upper living beings, unsupervised learning takes place through the interaction of the afferent (sensory) neuronal system and the efferent (motor) neuronal system. Although from a functional point of view there are no substantial differences, this interaction takes place at two levels, as shown in Figure 4:

  • The interaction with the inanimate environment.
  • Interaction with other living beings, especially of the same species.

The first level of interaction provides knowledge about physical reality. On the other contrary, the second level of interaction allows the establishment of survival habits and, above all, social habits. In the case of humans, this level acquires great importance and complexity, since from it emerge concepts such as morality and ethics, as well as the capacity to accumulate and transmit knowledge from generation to generation.

Figure 4. Structure of unsupervised learning

Consequently, unsupervised learning is based on the recursion of afferent and efferent systems. This means that unlike the models used in Deep Learning, which are unidirectional, unsupervised AI systems require the implementation of two independent systems. An afferent system that produces a response from a stimulus and an efferent system that, based on the response, corrects the behavior of the afferent system by means of a reinforcement technique.

What is the foundation of consciousness?

Two fundamental aspects can be deduced from the development of AI:

  • The learning capability of algorithms.
  • The need for afferent and efferent structures to support unsupervised learning.

On the other hand, it is known that traumatic processes in the brain or pathologies associated with aging can produce changes in personality and conscious perception.  This clearly indicates that these functions are located in the brain and supported by neural tissue.

But it is necessary to rely on anthropology to have a more precise idea of what are the foundations of consciousness and how it has developed in human beings. Thus, a direct correlation can be observed between the cranial capacity of a hominid species and its abilities, social organization, spirituality and, above all, in the abstract perception of the surrounding world. This correlation is clearly determined by the size of the neocortex and can be observed to a lesser extent in other species, such as primates, showing a capacity for emotional pain, a structured social organization and a certain degree of abstract learning.

According to all of the above, it could be concluded that consciousness emerges from the learning capacity of the neural tissue and would be achieved as the structural complexity and functional resources of the brain acquire an appropriate level of development. But this leads directly to the scenario proposed by the Turing Test, in such a way that we would obtain a system with a complex behavior indistinguishable from a human, which does not provide any proof of the existence of consciousness. 

To understand this, we can ask how a human comes to the conclusion that all other humans are self-awareness. In reality, it has no argument to reach this conclusion, since at most it could check that they verify the Turing test. The human comes to the conclusion that other humans have consciousness by resemblance to itself. By introspection, a human is self-awareness and since the rest of the humans are similar to him it concludes that the rest of the humans are self-awareness.

Ultimately, the only answer that can be given to what is the basis of consciousness is the introspection mechanism of the brain itself. In the unsupervised learning scheme, the afferent and efferent mechanisms that allow the brain to interact with the outside world through the sensory and motor organs have been highlighted. However, to this model we must add another flow of information, as shown in Figure 5, which enhances learning and corresponds to the interconnection of neuronal structures of the brain that recursively establish the mechanisms of reasoning, imagination and, why not, consciousness.

Figure 5. Mechanism of reasoning and imagination.

This statement may seem radical, but if we meditate on it we will see that the only difference between imagination and consciousness is that the capacity of humans to identify themselves raises existential questions that are difficult to answer, but which from the point of view of information processing require the same resources as reasoning or imagination.

But how can this hypothesis be verified? One possible solution would be to build a system based on learning technologies that would confirm the hypothesis, but would this confirmation be accepted as true, or would it simply be decided that the system verifies the Turing Test?

[1]Stanford Encyclopedia of Philosophy, «Kant’s View of the Mind and Consciousness of Self,» 2020 Oct 8. [On line]. Available: https://plato.stanford.edu/entries/kant-mind/. [Last access: 2021 Jun 6].
[2]S. J. Russell y P. Norvig, Artificial Intelligence: A Modern Approach, Pearson, 2021.
[3]A. Turing, «Computing Machinery and Intelligence,» Mind, vol. LIX, nº 236, p. 433–60, 1950.
[4]C. C. Aggarwal, Neural Networks and Deep Learning, Springer, 2018.
[5]L. M. Adleman, «Molecular computation of solutions to combinatorial problems,» Science, vol. 266, nº 5187, pp. 1021-1024, 1994.
[6]E. R. Kandel, J. D. Koester, S. H. Mack y S. A. Siegelbaum, Principles of Neural Science, Macgraw Hill, 2021.
[7]F. Rosenblatt, «The Perceptron: A Probabilistic Model for Information Storage and Organization in the Brain,» Psychological Review, vol. 65, nº 6, pp. 386-408, 1958.
[8]D. Silver, T. Hubert y J. Schrittwieser, «DeepMind,» [On line]. Available: https://deepmind.com/blog/article/alphazero-shedding-new-light-grand-games-chess-shogi-and-go. [Last access: 2021 Jun 6].