George Mobus teaches computer science to undergraduate and graduate students at the Institute of Technology, Computing & Software Systems at the University of Washington, Tacoma.
His background is quite broad: He has a PhD in Computer Science, an MBA in Decision Science, and a baccalaureate in Zoology (with substantial coursework in math, chemistry, and oceanography) from UW Seattle. His academic focus has been Biology: Specifically, evolutionary, cognitive, neuro-psychology — how the brain works to produce the mind and how did it come about through evolution.
He blogs at Question Everything, where this piece was originally published.
Sapient Judgment Has Weaknesses
In my working papers on sapience I describe one of the components of sapience as judgment and provide a brief overview of what role it plays in wisdom. I briefly mentioned some weakness or limitations to ordinary human judgment in that work, but left it a little vague. In this paper I want to go into more detail about how judgment works in the making of decisions and especially what some of the remaining problems with it are with respect to the current state of sapience in Homo sapiens.
Sapient vs. Pre-sapient Judgment
Sapience involves the capacity to influence good decisions (and give good advice) by applying judgment to complex situations. Decision making processing is the main job of intelligence as indicated in the working papers. Decisions need to be made regarding what action or behavior to take given the situation in the immediate environment. While the actual decision processing looks more continuous in nature, this discretized version will hopefully help to illustrate what happens in the brain*. The central circle in the below diagram can be considered a decision node in a decision tree structure (actually more of a web structure than a tree). The job that the intelligence processor has is, given the situation in which the animal finds itself, the “state of the environment”, to make a decision on which of many actions to take.
Figure 1. Sapient decision making depends on tacit memory models that influence intelligent decision making. See text for description.
Immediately surrounding the current decision point is additional information stored in working memory forming the context of the situation, how it came to be, factors that relate directly with the decision to be made. This context along with the current state that activated this particular decision point are fed into the intelligence processor’s causal model of the world. This model has been learned from past experience and represents the best estimate of what cause (from a selected action) and effect (on the future state of the environment as a result) to expect. The model actually contains the decision web along with several background influences such as affective valence marking that helps weight decisions when emotional considerations are operative (curved green arrow).
The learning component of intelligence is responsible for monitoring the actual outcomes of decisions vis-à-vis and building or refining the model over time. Then it uses the model to make selections for actions. The selections are always provisional or heuristic in nature. Intelligence processing is prone to several kinds of errors that might cause a misstep. Also, though not shown here, the creative function of the brain might intercede to suggest a different selection just in case it might lead to a better outcome and that could be the basis for modifying the model.
Some situations have strong survival aspects that are best handled by innate behaviors selected by the limbic system. For example, our innate reaction to a sudden loud noise (freezing, generally in a crouched position) is mediated by a limbic signal that overrides any learned behavior selection in favor of the innate reaction. It depends on the strength of the innate response selection criteria. Some forms of innate behavior can be subsequently overridden by learned behavior if the action selection of intelligence can overcome it**.
Finally we come to the role of judgment in affecting the decision processing of intelligence. In one way we might consider tacit knowledge as providing an extended context, in time and space, background that has been developed and stored in unconscious memory. This is a much larger and more generalized model of situations similar to the current situation plus a generalized history of outcomes learned from past experiences. This tacit memory model gets refined over time and multiple similar situations. It provides a richer set of influences on the intelligent decision making without explicitly entering into the consciousness. These are what we call judgments. Even more primitive mammals can make some small scale kinds of judgments as the cortical areas of the brain where these are made arose relatively early in mammalian evolution.
Pre-sapient mammals through primates, prior to the genus Homo, in general, have levels of judgment that increases with their position in the phylogenetic tree. If we think of more intelligent animals as higher on that tree, then we will find that their capacity to process judgments to augment intelligence is correspondingly greater as well.
When we get to Homo we see a great expansion of judgment capacity commensurate with their increased intelligence (reflected in brain size and expansion of the prefrontal cortex). This expansion involves having greater capacity for much more complex tacit memory models and especially models of social systems, moral considerations, and systems perspective. Then when we get to Homo sapiens we see the development of a full-blown capacity for strategic thinking, which, added to the capacity for judgment is what gives us that which I have labeled as sapience. The expansion here is not just quantitative but qualitative as well. It was a leap in capacity, and an integration of the four components of sapience (see working papers) that produced our much more advanced ability to make judgments covering a far greater range of time and space. In that expanded range of time and space our tacit memories could incorporate far more situation experiences along with molding of the models according to many outcomes. In terms of the above figure, think of the extended tacit knowledge block as being many times larger and its contents many times more complex.
As I have written before regarding the special role of human consciousness and its relation to sapience (see Human Consciousness and Sapience) another aspect of mentation that makes humans much different from prior hominids is the aspect of reflective consciousness, or what I called second and a half-order consciousness. This is where we are not only aware of the world and ourselves, but aware of our awareness. This kind of consciousness results in a form of self-observation of our capacity to make judgments, or reflective judgment. We can ask ourselves, mentally, if our judgments were good, bad, or indifferent. We can think consciously about doing better in the future. But most people do not go quite that far (it is the realm of philosophy that few venture into). Instead they are content with the notion that judgments come from some unconscious part of their brain they call intuition.
Most human beings do not really think about where their judgments come from. They just pop up and influence decisions in complex, usually social, contexts. Some people pay attention to these mysterious (gut) feelings and some even rely on them to make decisions because they have come to trust them.
The reference to ‘gut’ feelings as intuition refers to the limbic or affective influence on decision processing shown in the above figure. All of us have some kind of emotional marker on our past decisions (stored in the models) based on how good or bad the outcomes were. These emotional markers actually do activate the emotional state (not shown in the diagram) which sometimes does have an impact on our visceral state (our gut feeling).
For others, but I suspect in the minority, the intuitions influencing our decisions comes more from our tacit (learned) memory models than from our emotional responses. When they do, we have a special opportunity to reflect on our intuitive side. When we detect that our decision is being influenced by a tacit memory model, it is possible, in some circumstances, to reflect on where such influence is coming from and even examine the source consciously. Such reflection may help focus our intelligent learning on issues that are important to our tacit models such that these models are improved over time. One straightforward approach to this is that our conscious decision to explore ideas related to our judgments helps encode relevant information for subsequent further subconscious processing into tacit memory. Thus the philosophical musings of Socrates, “The unexamined life is not worth living.” It is people who do actively reflect on their own judgments that obtain some measure of wisdom in their lives. Alas, the rest do not.
But that isn’t the only problem with the sapience of or species. Sapient judgment evolved, after all, from more primitive kinds of judgment. And those were more strongly influenced by lower level (e.g. limbic) brain functions than we would like to admit. These low level influences are revealed, however, by the presence of many decision biases and prejudices that can be detected and even measured in some cases. The continued influence of these seriously impedes the ability for sapience to have a stronger role in guiding decision processing for most human beings. Let’s examine a few to show what I mean.
Biases and Prejudices
There has been an explosion of psychology research in the past few decades regarding innate human biases and the tendency of learned prejudices to strongly affect our decisions and behaviors. I will list a few of these here and attempt to explain how the get in the way of good judgment, and how the development of stronger sapience might do more to disenfranchise their influence. They are major blockages to the acquisition of wisdom and because they are so operative in the majority of humans, one of the main pieces of evidence I submit for why the sapience of Homo sapiens is not enough to be adaptive to the modern complex world we have created from our inventive cleverness.
Causal Modeling BiasOur brains are wired to find causal explanations for everything. Even when no actual causality may be operative. We are pattern seekers to the extent that we will even see patterns in completely random distributions of things in time and space.
This actually makes sense in a relatively stationary world where actual causal patterns could be relied upon to allow an animal to make predictions about what would happen (effect) given a particular kind of event (cause). In fact, our propensity to find causal relations is built right into the very neurons that make up our brains. Most of our learning comes from the fact that engrams (memory traces) are encoded based on cue stimulus, reinforcement (physiologically meaningful) subsequent stimuli, and good or bad outcomes (see my research paper: Learning and Representing Causal Inferences in Neural Networks). When reinforcement comes after a cue stimulus on a regular basis, the neural network (brain) will encode the cue as meaningfully causing the state that is relevant to the reinforcement. For example, when the sight of a particular fruit is followed by a savory taste/smell, and then followed much later by a satiation signal, that fruit is encoded as being food and something to be sought in the future. Eating it will make you less hungry.
But the world that we live in is so complex and causal relations so nonstationary, or even nonexistent, between multiple stimuli, that this tendency to encode any seeming temporal correlation (where event A precedes event B, usually), even from the most rudimentary experience, gets in our way badly. We have a tendency to imbue causal relations where none necessarily exist.
Stronger sapience involves having a stronger capacity to think systemically. In doing so it is natural to ask whether some phenomenon contains non-linear, hidden, relations (e.g. circular causality) and to observe more carefully when it seems that one thing occurring is related to another thing occurring. Weak systems thinking stops at asserting that things are connected (i.e. in some kind of network of relations) whereas strong systems thinking seeks to verify those connections and grasps the nature of strong versus weak interactions. Thus greater sapience will not be stuck looking for the simplest causal explanation and accept the weakest evidence (correlation) to form a model.
Likelihood Estimation BiasWe do not do statistics very well. Our ability to estimate probabilities of events or outcomes is pretty poor. This is related to our tendency to believe in our causal explanations (see confidence bias below); we do not think in terms of correlations but in terms of cause and effect. As a result we tend to make many mistakes in estimating probabilities given preceding information. We tend to ignore or filter out information that might improve our abilities. Sometimes we are simply ignorant of what is truly relevant information in making a prediction or assessment. Without reflective judgment this blockage to wise decision making gets worse rather than better with age.
With reflective judgment (stronger sapience) we do not ignore evidence that suggests that our likelihood estimates may have been wrong in the current circumstances. One of the key aspects found to be present in the wise (in the research) is an ability to live in an uncertain world, with ambiguity. This comes from an ability to not be committed to whatever estimates we have made, and to be open to learning more through experience in the hope (but not guarantee) of improving over time.
Over Confidence BiasThe part of our mind that Freud called the ‘ego’ seems to place greater confidence in our estimates than is warranted by the actual case. The mere personal possession of an idea or assessment boosts the mind’s confidence that it is right or true. Again, without self-reflection and self-honesty this tendency will get in the way of having good models — that is models that more closely correspond with reality — from which to intuit. When we ‘firmly believe’ our own version of reality regardless of what the evidence suggests, we can make pretty foolish judgments.
Religious or ideological beliefs fall under this bias. Holding a specific doctrine as an article of faith (without actual evidence, or very weak evidence) is bolstered by our seeming inability to reflect on our own confidence without prodding. And sometimes with prodding to become defensive about holding those beliefs. With higher sapience I suspect that an individual is comfortable with the notion that whatever they believe at the present time, might be wrong. They are open to reviewing their reasons for holding said beliefs rather than being so certain that what they believe is the truth. And this openness comes naturally rather than being learned.
Inability to Imagine Counterfactuals BiasRelated to over confidence is another weakness of the brain’s ability to generate counterfactual models in the first place. Numerous studies have shown that we humans are loathe to even consider alternatives to what we already believe. In fact, in order to more diligently assess possible (likely) outcomes we have to consider alternative models of what may happen. This is a function of imagination (creativity) that seems very weak in the majority of humans. Yet without the ability to generate and analyze counterfactual models, we have no ability to learn from actual experience. This is most evident in many peoples’ inability to learn from mistakes. What they are good at is generating excuses or rationalizations that let them off the hook for responsibility for errors of judgment.
Higher sapience would allow a person to let the imagination go with respect to generating alternative scenarios and find counterfactual conditions for evaluation.
Imagined Agency BiasAnother bias that is related to our causal modeling bias is the way in which we tend to see some intentional act behind many events even when they are completely stochastic. Indeed the beginnings of self-awareness have left humans believing that there is some kind of autonomous agency involved with all events. We have ourselves as a model. To us, to our consciousness, it seems as if we have intentions that our bodies then carry out. We have, in our view, free will. We also observe other humans doing what they seem to want to do (sometimes regardless of our wishes). This leads us to suspect that there are other agents behind all phenomena. Before the age of science, the belief in spirits animating not just living things but the dynamics of the whole world was common. Today many of our kind still hold out that there is an unseen world and one or more invisible and super powerful agents manipulating the world for their own purposes.
The combination of strategic and systems thinking in sapience, if strong enough, can lead to a better understanding of motivation of phenomena in the world and thus diminish the need to see agency behind dynamic phenomena. For example, I suspect Charles Darwin was a super sapient (not the only one, of course) who was able to grasp the essential nature of evolution. He saw the world through systemic eyes and had a sense of the hierarchical nature of control through competition and coordination. He could see that natural selection provided a strategic “plan” for the living world without the need for a designing and animating agent.
Illusion of Control BiasThe last bias I will discuss is related to the agency bias. It is an illusion that persists in the minds of most people that they (or someone) is in control of the situation. This is strongest in ourselves where our consciousness gives us the sense of free will and willing our selves to act in the world. But research has shown that our actions are often initiated from subconscious decisions and our conscious mind only becomes aware retrospectively, yet maintains the illusion of being the decider.
It is the models of the world that we have built up over time that are ‘in control’. We are our models. One can argue that it is a matter of semantics that if so, then we are still in control. But unfortunately the ‘I’ that seems to be in control, the model of the self, is the result of the interplay of inherited personality propensities and cleverness, and the social milieu that we grow up in. The ‘I’ is not some native spirit (the ghost in the machine) that pushes the buttons and pulls the levers. It is a subconscious model of the self resulting from many complex factors that develops in the course of a lifetime. It is true that this model is the strategic arbiter in our lives. It may be a good model, corresponding well with reality (knowing ones self) or it may be a faulty one in which we delude ourselves about ourselves.
Higher sapience would help here in the sense that it would provide an individual with a better capacity to build a more veridical model of the self as she ages and experiences life. That model, in turn, will have a better grasp of what needs to be decided for themselves and others. They would be better able to deal with uncertainty and ambiguity (as is reported in the literature on wisdom). They would be able to offer good advice without asserting that they were right or should control the situation and others.
Prejudice TemplatesHumans, like all social mammals (and birds) have a built in model of the world of conspecifics based on we (my tribe) and them (other tribes). This dichotomous mental model actually serves a purpose in strengthening the social bonds among the we and decreases the likelihood of fraternization (except to exchange genes) with the them. In a social animal, having a cooperating group that is competing with other groups for resources this has a strong selective advantage. And so we humans have inherited this template and it forms the basis of learning prejudices.
With all of the biases just mentioned (and more not mentioned), it is easy to acquire models of attributes of we and them that are not, in fact, reality. In a late Pleistocene world this probably was to our evolutionary advantage. But in the modern globalized world it leads to some pretty pathetic situations and a great deal of misery for too many people.
Given stronger sapience would reduce the effects of the aforementioned biases by constantly and consistently engaging in self-reflection it should also lead to a great reduction in the effects of prejudices. This can happen in two ways. The individual may simply never suffer from building a prejudicial model based on her socialization (learning prejudices from her society) and/or she may be successful at overcoming what prejudices she has by examining the basis of them and either quelling or unlearning them.
The Consequences of Weak Sapience
Here we sit today in a world that is overly complex and overly dynamic relative to the average person’s capacity to deal with information overload. As a result most people never develop much in the way of wisdom about that complex, dynamic world. One has to incorporate the information one receives into the structure of their tacit memory models in order to build veridical models of the world. But if one’s native processing capacity is limited then the information will truly overload the circuits and be lost. What aggravates this is the remaining built-in bias tendencies and the propensity to learn prejudices that further prevent any meaningful learning.
We have to muster all of our cleverness to try and understand our own failings in the present environment. This is not an indictment of humanity as if we did something wrong by being only weakly sapient on average. We are what we are. But fortunately we do have sufficient cleverness to be able to comprehend what we are and try to do something about it. We invented science through cleverness and minimal sapience. Science and math have helped us overcome many of the biases and prejudices that have been noted here. As individuals each of us still suffers to some degree the bane of these biases. We can’t change that. But collectively in an endeavor like science we can mitigate the effects of those biases and learn to understand better models of how the world works than would be possible through individual learning. So we have to call upon that capability here. We need to use science to better understand our failings in judgment and sapience in general. Then with the help of what science tells us, we might yet find a way to become collectively more sapient, even if individually still somewhat foolish.
* The issue of continuous vs. discrete has always been a problem. Computer scientists (my day job) insist that everything is discrete at some level of analysis. Biologists, on the other hand, insist that physiology and brain functions are continuous. Having played in both arenas I am of the opinion that they are both right to some degree. While to our perceptions brain functions might appear continuous and flowing from one state to another, if we look closely enough and with a fine enough temporal resolution I believe we will see that there are not an infinite number of transition states as would be required by continuum theory. Rather there are discrete, finite transition states between any two metastable states. The latter are the ones that maintain long enough for perceptual observation.
** One aspect of sapience is the capacity for the prefrontal cortex to dampen the signals from the fear and negative emotion response areas of the limbic system.
Question Everything: When what is happening in your world doesn’t make sense, when it doesn’t conform to your beliefs about how things should work, it’s time to ask hard questions.
What's been said:Discussions found on the web: