Entropy of the human soul

October 13, 2011 § 2 Comments

“Nyman, one day you will be a scientist” said my physics teacher Rafael Laurema, when I was sixteen and was presenting my solution to a physics problem at the blackboard, in front of my class. My school had the questionable (and unfair) reputation of being the worst one in Helsinki and the idea of a science career was reserved for those in better schools. No wonder that I still remember vividly his kind and encouraging remark.

“We are not mice,” said my friend and colleague Jerry White, almost fifty years later, over one of our conversations at Stanford on how to understand the essence of human behavior and especially how to promote a good life for those in trouble or for the elderly who are often seen predominantly as medical or sociological curiosities.

Predicting the life in a cage?

There is an inbuilt, profound conflict in the psychological theories of human character and individuality: on one hand the theories try to frame the concepts of “a person” and “an individual” as invariant human properties that carry over lifetime and that have predictive value for the individual life and behavior. On the other hand, most of us, if not all share the very basic desire and tendency for search, growth and creativity that makes us to choose, learn, cherish, to be cared for, change direction, interpret the world, invent and to adopt unforeseen tools.

These contrasting perspectives to the destiny of the individual life will not meet as long as static models and statistical generalizations dominate the psychological theories of development, intelligence and personality. These theories have become the nourishment for the frequent astonishment in facing unexpected behavior, in good and especially in bad. When someone breaks out of this box of simple predictability, like Steve Jobs apparently did, the public community is totally amazed and tries to find dramatic explanations to such an individual deviance. The same happens with evil deeds: remember for example, the profiling of the Norwegian murderer and the Finnish school murderers. The fact remains that we do not know which incidences in their near and far history or which specific personal factors in their lives led to their choices.

A common hope is to find a simple, invariant individual description, a personality model or profile that could be used for predicting behavior and experiences over the lifetime. At present, the static concepts that are used by various tests to characterize our personalities or intelligence offer the illusory promise that we can understand why certain behavior occurs, even in very complex circumstances.

Some researchers forget the individual characteristics totally. For example, quite recently we could read about the study where the researchers claimed that multivitamins actually increase the death risk for older women (http://www.businessweek.com/news/2011-10-10/vitamins-tied-to-higher-death-rates-in-older-women-in-study.html). Readers of this news are surely worried and even more so because it is impossible to know what such average-based results mean to each individual. Individual differences do not worry these trouble researchers.

Interestingly, in criminal investigations, the term “profiling” and behavior pattern analysis have been used for a wider spectrum and richer analysis of individual behavior. This is no surprise, because the criminal investigations deal with single behavioral acts that individuals perform, not with their statistical generalizations and average predictions or profiles, which cannot explain individual deeds.

A static model of human nature and personality discards what is essential and unavoidable in life: its complexity, interactivity, adaptation, and ill-defined character. The fact that a theory of personality or intelligence, for example, can have a statistical, average predictive value for simple behaviors does not mean that it would explain what an individual chooses to do, to sense and experience in a particular life situation.

Information theory for the soul

Why not accept the extreme complexity of individual life and to build the psychological theories of the individual on that? It is possible to see the concepts of  “character”, “personality” or “individual” as products of chaotic and complex social, biological, historical and situational forces that guide our behavior. According to this view, the behavior may appear systematic, from a simple static perspective but it is actually a product of a multitude of unknown, interacting, and simultaneous forces that can originate from the present or past contexts of the individual.

As an individual example, I have never been asked the simple question “Has anyone ever told you something that has had a significant impact on your choices in life?” But if someone had asked it, I would have immediately informed him of my physics teacher and of a couple of other persons significant to me. They have been an indispensable part of my life trajectory and they are not “background factors” that should be averaged out. We all have a significant personal history and hence, the world that we come to face is never the same for all of us. The question is, how is this reflected in what we choose to do in our lives?

The optimistic side of this complex view is that it becomes possible to assume and accept that certain personal decisions are not inevitable and that the chaotic forces guiding our behavior can also be the creative sources for choice that open up a multitude of alternative ways to behave in a specific situation. Without such forces, arts and creative sciences would be lifeless.

The grace of entropy

One way to characterize a behaving system – or individual behavior – is to use the concept of information entropy (cf. http://en.wikipedia.org/wiki/Entropy). In information theory entropy characterizes the amount of randomness in a behaving a system. Entropy takes us to the world of disorder and in this sense it can be difficult to accept for psychological sciences and thinking that aim at simple and economic solutions to complex problems.

Entropy is a measure of disorder and chaos: high entropy means that the exact and detailed behavior of the observed system cannot be predicted with a high accuracy. Statistically we may be able to model it but the actual behavior, its next state or act, for example, is impossible to predict. In this sense, entropy encapsulates perfectly the challenges that psychologists and psychiatrists meet when diagnosing and measuring human nature in order to predict or understand their individual behavior in specific circumstances.

A lesson from the ant

In his book “The Sciences of the Artificial” Herbert Simon admires the apparent complexity of an ant’s path. One could be tempted to think that such a complicated path is produced by a rather complex or even an intelligent system. However, the teaching is that we should accept the complexity of the “life output” as an essential fact of life, a natural ingredient of it.  But a
complex pattern need not be the output of an overly complex organism or machine.

Many psychologists and even some shortsighted epidemiologists and proponents of gene theory have not received the ant’s lesson and by using static measures or concepts to describe behavior, illness, intelligence or personality they assume that a sufficient criteria for a successful psychological theory is its ability to accurately predict very simple behavioral or functional phenomena.  They forget that in order to predict significant life events, the predictions must be complex, perhaps non-separable in nature. This is why it so difficult to accurately tell what role a specific gene or an intelligence level will have in our future lives.

Why not look at the individual, her choices and experiences as the most complex phenomenon we can imagine and accept it as the starting point?  Doing so we would explicitly pay respect to human life in all its forms. The fact that behavior appears simple, sometimes straightforward and predictable is not against the complexity, it only means that the perception of this simplicity is the product of how we perceive complex phenomena. When a complex process is mapped on our limited –dimensional frame of reference we, as humans and scientists can recognize only simple behaviors in it. Sometimes we need this perspective, for very practical purposes. But in order to understand human nature we must step outside this frame, and as scientists, it is possible to extend and transform this perspective.

Simple case of complexity

There are ways to observe human behavior, with these thoughts in mind. Some years ago, a colleague of mine, Dr. Jari Takatalo started studying how people move in a virtual 3D (CAVE) environment where they were given a simple object search task. The aim of the study was to identify individual movement patterns that would be characteristic to different test subjects. In other words, we were searching for personal “fingerprints” in their motion patterns of search behavior (Särkelä et al).

Naturally it would have been possible to use various statistics to describe the average movement patterns of our sample in these tasks but then our colleague and friend Dr. Patrick May (Aalto University) suggested the use of entropy measures to characterize these recorded movement trails. A simple way to measure entropy would have been to compute it for each subject, from the (x,y) locations during his/her search task in the CAVE. Entropy would then be an expression of the randomness or complexity of the movement pathway and high entropy would be an indication of “an ant’s trail”, a totally random pattern of movement in the (x, y) space.

Then we realized that the entropy of more complex behavior elements could be more interesting: why not use the velocity, acceleration, turning, or other more complex, higher -level aspects of behavior as input to the entropy calculations and to compute entropy for them? For example, we could obtain a measure for the randomness of the turnings of the subjects. In other words, we wanted to see if we could find the most basic elements underlying complex behavior, that is, those elements on the basis of which our subjects made their choices in deciding how to move, from one point to another.

But this kind of thinking need not be limited to such simple motor behavior analysis; we can extend it to more symbolic and complex levels of life and behavior. For example, in my own life, I know that these two utterances, ““Nyman, one day you will be a scientist” and “We are not mice,” have significantly guided my behavior and choices and will continue to do so, beyond my “personality or intelligence profile”.  But this is just surface and we should include into these considerations the significant events of our experiential and emotional history, instances of concept learning, habit formation patterns, individual reaction tendencies, and many others. The future theories in psychology will learn to accept this level of complexity and to deal with it theoretically. All the tools are available already.

I have no extensive scientific data to defend this view, but it is an inspiring thought experiment. There are already ongoing experiments in this type of data collection but the analysis methodologies remain to be developed for individual life time data collection (cf. the Quantified Self community http://quantifiedself.com/). But perhaps in a very near future we can better recognize holistic individual behavior patterns and also learn to better predict or at least to understand individual behavior, in the future good and bad.

A sceptic could argue that a rich information theoretical  analysis leads to excessive complexity and overload of information and that there is no way to collect such an immense mass of individual data. But think again: the development of ict technologies is already now providing access for extensive individual data collection and actually, most applications like Facebook simply throw away valuable individual behavior data. A totally different question is the ethics of such data collection. But for willing individuals, I believe this could offer a better future in the hands of psychologists and psychiatrists who test them with their present static tests and lack of massive behavioral data of us.

If we accept this perspective then the next question is, what are the most basic and valuable elements of behavior that should constitute the input to the entropy analysis of individual life?  And then of course, there are other advanced methods to approach such complex phenomena as individual behavior and choices. Psychologists should be the masters of them.

Where Am I?

You are currently viewing the archives for October, 2011 at Gote Nyman's (gotepoem) Blog.