“When it is not in our power to determine what is true, we ought to act according to what is most probable.”
Discourse on Method, Rene Descartes 1637
We typically think of belief as something only conscious beings are capable of; Mary believes achieving her degree will improve her future, Fred believes in pyramid power. A wider definition of belief recognizes that it can be extended to assist in explaining any behavior that can be characterized as goal seeking. Why did your dog Fido follow you into the kitchen? Fido went to the kitchen because he ‘believed’ he was about to be fed. Behaviorists turn in their graves and a few philosophers are quick to accuse those using the term belief in this way of anthropomorphism and teleology. They say indulging in such modeling projects the human experience out onto events in the world that are not really there, like a primitive animist. Worse, the critics see in such characterizations a throw back to the teleology of Aristotle’s physics in which everything had a final purpose, in which rocks fell because they wanted to rest on the ground. It is true that the strict semantic meaning of belief does not apply to purely physiological phenomenon. Of course just where belief does arise is a tantalizing question in light of the ‘astonishing hypothesis’ (Crick 1994) that all brain states arise from the activities of neurons. Are beliefs to be found in the individual neuron, a circuit of neurons, a web of interconnected circuits or some global orchestration beyond anything conceived in the neuroscience of the early 21st century? We are not going to start with such perplexing considerations. As we saw with Fido there is utility and explanatory power to be gained in a wider interpretation of the term. To properly ground just how fundamental belief is to the human experience we start by exploring just how far it can be used as an explanatory vehicle.
As we seek to explain the movements of rocks and mountains, stars and water we can say they seek some goal but doing so adds nothing to our understanding of physics. Things are very different when we turn to consider the biological world. Here even the simplest of organisms can be said to have a goal. A virus can be said to seek its propagation by infecting its host. A virus is the simplest multi-celled life form known, so much so that whether to define it as alive or not turns wholly on the various definitions of what it means to be alive. Unlike the case of rocks and water, postulating that the virus has a goal and seeks to achieve it does aid in our understanding of the behavior we observe. No one who talks this way mistakes the meaning of what is being said, they do not ascribe consciousness to the virus.
Goal seeking behavior is everywhere in biology. Consider the paramecium. Imagine you are looking down a microscope at this simple creature in a drop of pond water. It is a single celled ciliate protozoon, its cellular membrane covered with tiny hair-like cilia that allow it to move. It is good at moving, able to traverse a distance from four to twelve times its body size every second. A six foot man moving twelve times his body size would cover seventy two feet in a second, a bit under a mile a minute. It moves so quickly microscopists often use a thickening agent in the water to slow it down and aid observation. Why does it move? It includes another set of cilia around its oral groove that allows it to consume food, it moves to eat. The movement observed will be a random walk but surprisingly if it encounters an obstacle it is able to reverse its direction. It has a calcium activated reverse gear. The process of changing from forward motion to reverse is controlled by voltage polarizations in which the leaking of ions is not perfectly smooth. The result is that when it again resumes its forward motion it will be in a slightly different direction. This single celled animal seems to avoid the obstacle.
Moving up the life complexity scale just a little, imagine gazing down the microscope at the most predominate life form on the planet, a bacterium. It too is a single celled organism and found in just about every habitat on earth. A millimeter of fresh water might contain a million of them, a gram of good soil 40 million. The bacterium will scuttle about in a random walk using their whip like flagella. In E. Coli a counter-clockwise rotation of the flagella causes forward motion, clockwise causes a tumble action changing their direction and together the processes produce the random walk – some time spent moving forward, a tumble to change direction followed by another spurt of forward motion. Food for a bacterium is usually a simple sugar. Imagine that the water drop under the microscope contains a solution of glucose. The concentration of the glucose will vary, some areas having a higher concentration than others producing a food gradient. Now the behavior of the bacteria changes through a process known as chemotaxis – the random walks become biased towards the area of highest concentration of glucose. They wander towards the food source. Equally amazing, they are able to move away from a source of poison. How can a mindless creature display such purposeful behavior? It has a goal and adapts its behavior to achieve it. How can this be? The molecular explanation is that the chemicals are sensed through transmembrane receptors which send signals to Che proteins which alter the tumbling frequency.
Paramecium and bacteria are single celled organisms that do not contain a nucleus, they are prokaryotes. They are too small to sense the chemical gradient directly. This changes when moving up the life complexity scale to those single celled organisms that do include a nucleus, the eukaryotic. The amoeba is a single celled organism with a nucleus. As we saw, the prokaryotes scan their environment with constant swimming in their run and tumble fashion. In contrast an amoeba is large enough to detect the gradient directly; some amoebas are the size of a grape! They detect the differences in solute concentrations using receptors along its membrane. Instead of a biased random walk as displayed by the paramecium the microscopists observe a motion more directly towards the highest concentration of food. A dynamic polarization of receptors results in unmistakable goal seeking behavior. This single celled organism senses food and darts straight towards it.
In seeking to understand this behavior we turn to the microbiology involved but there is another level of explanation that arises when all the molecular details are orchestrated into a complete picture. This is where it makes sense to speak of goals and the algorithms or strategies required to successfully obtain them. In the example of the amoeba’s food gradient the algorithm is – if there is a higher concentration of food to the north east follow it because it is highly probable that the saturation will increase further in that direction. This level of explanation is not tacked on, ephemeral to the concerns of what is really going on. This level allows us to understand, predict and control the behavior in ways inaccessible without it. It is also testable independent of the molecular level, though of course any such tests depend on those intricate molecular interactions. Allowing this algorithmic level of explanation and using the Bayesian equation we can model the behavior of the amoeba on a computer. There is no interview process that allows us to question its prior belief in which direction a food source might be found as we did in the little cognitive test in the last chapter. Instead we estimate an objective prior, basically a model of complete ignorance. This allows the process to get started. We input the strength of the food saturation at the starting position, turn the equation crank and out comes the posterior probability it uses to move in a chosen direction. This directed motion in turn provides more chemical evidence which is used as new input, the previous posterior becomes the new prior, the crank is turned again and the process continues. Bayesian equations can often be applied sequentially in this fashion.
There are some who would deny the correctness of this higher level explanation that concerns itself with goal seeking. There are three typical objections. The first is physiological; ascribing intentionality to something like a single cell is just not right, it doesn’t even have a nervous system for heaven’s sake. Other biologists disagree and join Schopenhauer in describing something like ‘The World as Will and Representation’ (Schopenhauer 1958) seeing everywhere in living things a will to survive. These are interesting questions, important but not wholly pertinent in this context. The higher, algorithmic level is postulated for its utility, it is not necessarily a statement about how things really are in-themselves but an as-if that can be justified because it is found to be useful.
The other two objections arise from particular views of science. In the first a strong empiricist position asserts that the only concepts with scientific validity are those that arise from what can be directly observed. This program of the logical positivists failed in practice and the modern philosophy of science on the whole dismisses its overly restrictive claims. Pure behaviorism in psychology is an example of the positivist approach. This attempt to build a science of human psychology without reference to thought or emotion, belief or goals has been rather thoroughly discredited by modern cognitive science. The second view of science that objects to belief used as an explanation is nuanced and many consider it valid even today. For these thinkers only a reduction of all phenomena to the laws of physics is considered the proper work of science. It is a nuanced position because in some forms it is insisting that all things must have a material or energetic basis. The scientific method depends on a methodological naturalism that agrees with this in principal. Those that object to an explanatory use of a concept like belief in the examination of the behavior of living things go beyond this. They ascribe to what Daniel Dennett has accurately labeled ‘greedy reductionism.’ (Dennett 1995) For the most part it is a caricature of even the hardest of the hard scientists. There are concepts capturing causes and effects on levels of organization and complexity above the atomic that are every bit as real as the atoms themselves. Temperature is a classic example. It is a real phenomenon that can be measured and used to predict and control events through the laws of thermodynamics, yet temperature only exists at the level of aggregated molecular movement. It requires the material substrate as indeed all things do, but does not exist at the level of the individual atom and its isolated reactions. A greedy reductionist, to be consistent, would need to deny the reality of the phenomenon of temperature. There will be more to say about methodological naturalism and the importance of independent levels of explanation in a later chapter devoted to an examination of the scientific method and how Bayesian confirmation theory works.
Biological systems have evolved numerous adaptations in the long history of life on earth. The diversity of the “endless forms most beautiful” that grace our planet is breath taking. Natural selection has rewarded those forms that adapted to the challenges of the environment better than their competitors with successful reproduction. Generation after generation the combination of selection and genetics has worked its way. Each organism finds itself in a unique environment in which to implement the algorithms that maintain its survival. Each living organism embodies the goal of reproduction among all the others required for the ongoing day to day business of surviving. It is a mistaken understanding of evolution to say that an organism has evolved towards any particular goal. However, it is equally mistaken to take this as saying the products of evolution do not have goals. Evolution is guided by iron constraints. Mutations that lead away from the reality of the environmental challenges are ruthlessly culled. The complete relevant environment includes the internal milieu of cellular metabolism, the external physics and chemistry in which the biological processes unfold and the society of other organisms including both those like and unlike oneself. All of these aspects of the real world present regularities forming the fitness landscape on which the competition and cooperation of life unfolds. That there are regularities allows adaptations to incorporate them in the algorithms and strategies they develop. The environments present regularities, a useful term that captures the probable nature of most circumstances. The web of causes in all these environments is so complex most events important to survival cannot be said to occur inevitably, only that they are likely to occur more often than not.
Our amoeba moving to the north east does not know that direction will always lead to a higher concentration of food, only that it is the most likely direction to pursue given the environmental information it currently has. Life can be seen as a basic process for gathering relevant information and using it to control the choice of actions. It does so through a set of heuristics designed to improve the chances of survival and fitness. These heuristics necessarily involve an element of probability. The existential circumstance is wonderfully summed up in the title of Jacques Monod’s masterpiece, “Chance and Necessity: An Essay on the Natural Philosophy of Modern Biology” (Monod 1970).
This is the tapestry against which our considerations of the power of belief will unfold. The virus, bacteria and amoeba cannot change their beliefs; they cannot choose them or mold them. It is an open and interesting question to just what degree Fido can. A well trained dog given a command to stay will remain sitting and not approach the food bowl. The desire to eat remains, the goal to survive still operates but the behavior is modifiable. If Fido was starving… well the answer is undoubtedly contingent on each individual canine but it seems reasonable to assume that there is a point beyond which the command to sit and stay cannot go. These same primitive, fundamental, biological, goal seeking beliefs no doubt continue to operate in us as well. To what degree they are modifiable and how they are modified is a critical question. Undoubtedly some are unlikely to be any more penetrable by conscious influence than what we have seen in our distant relatives. In a sense life lives us and in the process belief acquires deep and powerful roots.