Scientific Brain Model Chapter I

 Chapter I - Intelligence is not a magical light bulb inside our brain; rather the brain learns plans of action for the achievement of the organism's goals.

In order to understand how the brain works, it is best to think of it as the government of the organism. It is then natural to say, that all important decisions are taken by our (conscious) mind. However, if we analyze a little deeper how the brain works, we will quickly realize, that in this context the word 'decisions' is rather misleading: 

Our mental constructs are nothing but the result of our experience with the environment. As a matter of fact, our mental constructs' raison d'etre is the formation of a model of the world outside, where - in order to optimize our interaction with our environment - our interpretation of how the world works is represented. It is therefore really hard (if not altogether impossible) to conceive anything foreign to everything we have so far been exposed to. For instance, as three-dimensional beings interacting in this world with other three-dimensional structures, it is really hard for us to conceive any four-dimensional shape. Since it is also our daily experience, that choices are always made by some decision-making agent; it is then likewise natural to assume, that this must be exactly what happens in the brain: intuitively, our senses gather information about the environment and elaborate a description of its current state. Once this has been accomplished, it is our consciousness' job to analyze this description and decide on a course of action. Not only does this follow the scheme of everything, that we have previously been exposed to; but it faithfully corresponds to our sheer sense of existence and very perception of life. Indeed, how often to we make remarks such as" "I see...", "I hear...", "I believe...", "I think...", "I like...", "I want...', etc.? We will even make statements such as: "The Egyptians decided to drop the Hieroglyphs System", instead of simply: "eventually other more efficient and practical writing methods became available and, with time, less and less people were taught    how to use hieroglyphs". Certainly, it would sound really strange to hear anybody saying: "the neurons in my brain reckon...". Yet, neuroscientific evidence strongly indicates that the latter reflects much better how our brains actually function. If there is any climactic discovery which any scientist could ever dream to achieve, that is undoubtedly the finding of the neural processes responsible for personal agency. Yet, as deeply, thoroughly and hardly as the brain has been examined, no evidence has ever been found of any group of neurons playing the role of any decision-making agent. In fact, it is not just that there is no evidence of any group of neurons taking decisions based on some description of the current state of the environment; but the sheer scheme of some agent analyzing any such description and making choices accordingly, is not in agreement with anything, that is known about how information is processed in the brain. 

* For instance, recognition is not accomplished by an intelligent agent analyzing the description of the environment generated by our senses.

 

To begin with, the evidence does not support the notion, according to which any such description of the surroundings is build, by those brain areas responsible for the processing of sensorial information. Eminent Psychology Professor Irving Biederman dedicated his entire career to the search of evidence in support of a model along these lines; but, as skilled as he certainly was elaborating his arguments, the data he presented remained far from conclusive. Moreover his 'Geon' (Generalized Cylinders) theory of visual cognition only addressed how human-made objects could be represented; but did not provide an account for natural structures.

* Action-oriented perception: our senses collect the information about perceptual stimuli, which our brain learns is relevant to perform the organism's actions.

 

Fact of the matter is the scientific evidence favors instead an 'action-oriented'-perception model such as the one proposed by Michael A. Arbib. The idea behind action-oriented perception is that we perceive our environment and the events within it, in terms of our ability to act. In other words, the processing of sensorial information is oriented towards signaling 'affordances' or actions the individual can afford to perform. For instance, the sight of a mug's handle elicits the activation of the offordance to grasp the mug by its handle. Albeit it can eventually get as sophisticated as necessary, the action-oriented model can be seen, in basic terms, as a direct mapping from perception to action.

* If our ultimate objectives are to stay alive and be happy; how do we accomplish these two contradicting desires? 

  After all - if the ultimate purpose of the mind is to optimize our interaction with our environment - the actual question to be resolved is what is the optimal course of action in any given situation. In other words, the real question is not, what is what we want; but how to achieve our ultimate goal in life. Indeed, there is little doubt that, for all of us, the ultimate goal in life is to survive and be happy. Yet, the fact that nobody seems to be able to consistently be happy - not even those, who have all the money in the world to buy everything they want -, indicates that we just do not know, how to achieve happiness. It therefore seems a reasonable approach to come up with a series of subgoals, which we hope, guess and sometimes even reckon will get us a step closer to the wildest of our dreams. In fact, as we consider what may help to get us happy, we begin to realize that, if the pursuit of happiness is so elusive is because the objectives, that we can think of, are usually in conflict one to another. Indeed, in order to survive we undoubtedly need to look after our own interest; however, likewise critical to our happiness is to be loved by the people around us. Now, as much as we may hate it, it is rather unlikely that other people will feel much love for us, if we exclusively care about ourselves and completely disregard everybody else's needs and desires. In general terms, in this really complex world we live in, we often do not know and it is not exactly straightforward to figure out how to accomplish most of our biggest objectives. Clearly, the difficulty does not lie in deciding what we want, but in finding out how to accomplish it.     

Divide-and-Conquer, however, is a general strategy that works particularly well for complex problems. Laying out the different subgoals, which need to be fulfilled, constitute then a course of action to achieve our global aim. Interestingly, any of our goals can be seen as a desired future state. Indeed, the fulfillment of any need or wish is nothing but a desired internal state. Moreover, it then does not seem too complicated to establish a link between an external state of the environment and our desired internal state. For instance, if we want to find some specific person in a crowd of people, spotting that person's face would represent the target external state. Similarly, for a frog looking for some lunch to fill its belly's desires, an exciting external state would certainly be the sighting of some juicy fly in its environs.

* Does the frog's brain have an intelligent agent responsible for choosing which of two flies to snack at, or is it just neurons competing for which of the two flies to snap at?

One of the main obstacles in brain research is the temptation to resort on incredibly deeply-rooted ideological preconceptions. For instance, if right from the outset we assume there is a soul responsible for our most profound thoughts and ideas, we are left with little else to explain. THe study of animal models is for this reason of even more benefit than what is already normally the case. In fact, - as much as we enjoy thinking that we are so much intellectually superior than any other animal - if we set aside all our prejudices, it is really hard to find any intellectual faculty, that lies beyond the potential of any other animal's brain. Yet, given we do not assume that easily the existence of an animal soul, we are forced to explain those intellectual abilities in some other way. Clearly, if a frog snaps at an unwary fly, which injudiciously chose to fly in its vicinity, the frog is certainly not making a conscious decision, but, surely, it is only reacting instinctively. Definitely, the stupid fly foolishly put itself in the frog's cross hairs, and the latter simply fired accordingly. Even if we consider a more complex scenario, where two flies fall in the frog's field of view (say one fly in the left visual field and the other in the right visual field), the frog will not need to be equipped with any sophisticated, fully-evolved (intelligent) decision-making consciousness, in order to be able to judiciously snap at the juiciest of the two flies. 

* Choices are made by means of competition between groups of neurons.

Indeed, we could think of a model such as Didday and Arbib's 'Maximum Selector', where frogs simply evolved to fire at whichever fly appears the fattest. As experimental observations reveal, neurons in the frog's tectum (the frog brain's equivalent to the primate's visual cortex) fire, whenever a specific preferred stimulus falls into the cell's visual field (receptive field) Neurons in this kind of sensory brain areas are always arranged topographically; so that we will find close to each other those neurons, whose receptive fields likewise lie close one to another. The presence of a fly somewhere in the frog's visual field, will then trigger a vigorous response from those neurons, whose receptive fields correspond to that region in the visual field, where the unwary fly chose to spend the last moments of its life. Indeed, the firing of these neurons in turn signal an 'affordance' for the frog to snap at such delicious snack. Now, not all is yet lost for the injudicious insect; but luck may have it that a second ill-advised fly enters the frog's visual field at such critical time, offering itself in sacrifice to save its buddy's life. The fool's heroic action will then trigger in turn a second affordance for the frog to fill its belly's desires with some other piece of brainless meat. At this point, we can tell that these definitely must be some truly stressful moments for our amphibian friend, completely torn between two such exquisite delicacies. 

Now, the 'action-oriented perception' framework does not preclude the existence of a intelligent, decision-making agent. Rather, it is still perfectly possible to imagine an agent performing some assessment of the different affordances available and selecting for execution, whichever is believed to be most appropriate. In fact, if our pal were a human, we would undoubtedly say that some intelligent, decision-making agent in its brain will, for the good or for the bad, eventually make a choice between the two items in today's menu. But, since it is only a frog, we will look for some more scientific answer. In such a case, the frog's agency is implemented as a competition between the two affordances: whichever fly is estimated to be the fattest, will become the frog's lunch for the day. In fact, all the evidence support the notion that there will be a competition between the neurons advocating for the execution of the first fly entering the frog's visual field from the left and the neurons advocating for the execution of the second fly coming to the rescue from the right. . In this context, the cumulative strength, with which each of these two groups of neurons fire, give a measure of the confidence in the fitness of each of the respective proposals. Then, in a similar fashion to a battle between two armies, whichever (overall) fires more energetically (if given sufficient time) will come out victorious and impose the next action to be carried out. Whether there is a decision-making agent or not, for all intends and purposes, a choice has been made.

Indeed, whether there is a intelligent, decision-making agent or not, the whole point of intelligent behavior is to optimize the system's interaction with its environment. That is, at any given time, based on the current perceived state of the environment, the system applies the knowledge it has acquired throughout its lifespan, in order to come up with the optimal course of action most likely to fulfill the system's needs and desires. Certainly, it is always possible to view the system's knowledge, together with its needs and desires, as the fundamental components of the system's intelligent, decision-making agent.