Confidence, Part III

In Confidence, Part II, the authors conclude that confidence is computed continuously, online, throughout the decision making process, thus lending support to models of the mind as a device that computes with probabilistic estimates and probability distributions.

 

The Embodied Mind

One such explanation is that of predictive processing/embodied mind. Andy Clark, Jacob Hohwy, and Karl Friston have all helped to weave together this concept. Our minds are blends of top down and bottom up processing where error messages and the effort to fix those errors makes it possible for us to engage the world. According to the embodied mind model, our minds do not just reside in our heads. Our bodies determine how we interact with the world and how we shape our world so that we can predict better. Our evolutionary limitations have much to do with how our minds work. One example provided by Andy Clark and Barbara is a robot without any brain imitating human walking nearly perfectly (video go to 2:40). Now how does this tie into confidence?  Confidence at a conscious level is the extent of our belief that our decisions are correct. But the same thing is going on as a fundamental part of perception and action. Estimating the certainty of our own prediction error signals of our own mental states and processes is as Clark notes:  “clearly a delicate and tricky business. For it is the prediction error signal that…gets to ‘carry the news’.”

Warning lights

Clark uses the example of the temperature warning light in your car being too sensitive and reporting slight fluctuations above some temperature. You believe that your car is overheating, but according to the garage personnel it is not. Moreover, the garage personnel have not seen the warning light and may have suspicions about you. So you find a temperature warning light warning light. Unfortunately, now we have just pushed the problem further back. If both lights flash, we now have to determine which carries the most reliable news. This does not help our confidence. So, as with many things, the balance between inference and action determines our response.

Sensory attenuation

In the paper “Active inference, sensory attenuation and illusions,” 412 Cogn Process (2013) 14:411–427, Harriet Brown, Rick A. Adams, Isabel Parees, Mark Edwards, and Karl Friston write that to make active inference/embodied mind or predictive processing work, there needs to be a way for top down or bottom up signals to prevail. In some sense, there needs to be a way to determine the winner–which you will listen to or more accurately, act on. In order for descending signals to prevail, the precision of the sensory prediction must be attenuated. This means that you must temporarily reduce your confidence in those sensory cues. This can only be done for intensity based signals like loudness or pressure and not for things like hearing frequency. The easy examples of sensory attenuation are like tickling oneself.

Thus, Clark explains, action requires a kind of targeted dis-attention where current sensory information is attenuated so as to allow descending predicted sensory states (proprioceptive) to determine movement. This might explain the experience of “choking” where we are engaged in a well practiced physical activity–say shooting a free throw in basketball or a two foot putt in golf. In such cases, deploying of deliberate attention to the movement seems to screw up the higher level proprioceptive predictions that would otherwise assure fluid movement and we miss.

Brown et al suggest that sensory attenuation is an interesting phenomenon partly because sensory attenuation is reduced in schizophrenia, or those at high risk of developing psychosis. In normal subjects, sensory attenuation is (negatively) correlated with the level of delusional beliefs. Less sensory attenuation means that the sensory perceptions of schizophrenics are more accurate than controls and—in the force-matching task—they perform better. A key symptom of schizophrenia is aberrant perception of agency, particularly the delusion that one’s actions are being controlled by others, suggesting the mechanisms that impair sensory attenuation in schizophrenia are intimately related to the perception of agency. These examples illustrate that being able to effectively turn on or off certain signals can make us more or less confident in our decisions or actions. Great athletes can turn off certain sensory information so they are confident and do not choke. Schizophrenics cannot turn off certain sensory information so they are confident others are controlling them.

Parallel Constraint Satisfaction

So with the prediction machine/embodied mind/prediction error minimization Clark and his colleagues have created an explanation of how perception and the mind might work. Interestingly this explanation is in line with the Brunswikian Lens Model and its more advanced offshoots such as the Parallel Constraint Satisfaction–Decision Making Model (Glockner, Betsch, Jekel) and alternatives.These descriptive models specify a general network structure and a flexible transformation function, thus allowing for precise predictions of choice, decision time, and confidence. At the same time, it achieves adaptivity through one free parameter that captures intra -and inter -individual differences in the sensitivity to (the distribution of) cue validities. Specifically, individuals may differ in how they translate information about the world into their mental representation of the decision task. The idea that a significant amount of our personality or cognitive style including confidence or perceived intelligence might be based on our differences in a single parameter may seem crazy, but aggregation modeling shows us the complexity that can be generated by a simple rule.

So we may have individual  tendencies in our confidence of cue validities. As we learn and update cue validities and weights, we may differ in the speed at which we make such changes. This may create tendencies for some of us to decide that we need more information more quickly and change from the automatic system to the deliberate more often and more quickly. Of course, depending on the learning environment–wicked or friendly, our individual tendencies may help us make good decisions or not. So the supremely confident decision maker going with his gut may do better in a wicked learning environment, because he is not trying to learn. There are environments in which each of our confidence styles might make better decisions. We just have to be lucky enough to find ourselves there. At the same time, if we are so confident that we always go with our gut and never change our cue validities or weights, that is, we never deliberate, seek new information, and learn, our decision making success will definitely degrade over time.

 

 

 

 

 

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.