Can a perception-cognition-action system function without values?
Opinions from the field
Traditionally in discussions of the mind, people have believed that logic can function well only in the absence of emotion, that emotion interferes with reasoning ability, and people would really be much better off without it. Many philosophers and scientists, even today, are dubious about the role of emotion in the mind; Pfeifer and Scheier note that "most people think that higher mammals, in particular apes and dolphins, but also dogs and cats, have emotions. Whether emotions should be considered an essential feature of intelligent beings, however, is debatable" (Pfeifer and Scheier 1999, p. 11). Others, however, have recently a different point of view; Henry Plotkin, for instance, says that "emotions are there in order to tell us what to think about; our hearts not only try to rule our heads, but should perhaps be allowed to do so" (Plotkin 1993, p. 207)
Emotions as values
The issue to be dealt with here is whether a perception-cognition-action system can function without values, but the main topic thus far has been emotions. This is because emotions are a biological system's values. They indicate what to perceive as good, what to perceive as bad, and what to perceive as somewhere in between. In a biological being, emotions are instantiated via certain brain states as well as heart rate, muscle tension, sweating, and other bodily responses. Antonio Damasio (1994) notes that it is also possible to have "as if" emotional states, in which the brain conjures up the emotional state without making the body go through the energy-inefficient process of actually instantiating the emotion.
Within the brain, emotion-related pathways are generally found within the frontal lobe. This is where pathways for different types of knowledge come together and continue on to the motor control areas of the brain (Broude 2003a). If damage to this area -- which includes the ventromedial, dorsolateral, or somatosensory areas, and the amygdala -- occurs, it results in problems with the processing of emotion. Damasio describes his frontal lobe patients as retaining skills in memory, mathematics, formal reasoning, and other non-emotional processes but losing social functioning, which turns out to be a significant problem.
Come on, you can't really believe that emotions always help.
It is obvious that emotion is not helpful for every situation in life. Math problems, for instance, do not require much emotion to arrive at a correct answer. Sometimes, in fact, emotion can actually interfere with reasoning. Even Damasio, who wholeheartedly argues for the usefulness of emotion, admits, "Although I believe a body-based mechanism is needed to assist 'cool' reason, it is also true that some of those body-based signals can impair the quality of reasoning" (Damasio 1994, p. 191). He cites a patient with damage to the frontal lobe who saw the woman in the car ahead of him skid on a patch of ice, panic, and end up in a ditch. The patient, on the other hand, drove smoothly along and got safely across. Many people would have been affected by their shock at seeing the woman in front of them, or by their own panic if they felt their own car slipping; but because this patient did not process emotion normally, he simply recalled the proper procedure to deal with a slippery road rather than panicking and slamming on the brakes.
As inviting as it may be to assume from instances like this that logic must work without emotion all the time, this simply is not the case. In many situations one encounters throughout everyday life, there are too many options to be able to go over each one in a logical way. Damasio cites an instance in which he asked the same patient previously mentioned to choose between two dates for their next appointment. "Just as calmly as he had driven over the ice and recounted that episode,” Damasio says, “he was now walking us through a tiresome cost-benefit analysis, an endless outlining and fruitless comparison of options and possible consequences" (Damasio 1994, p. 193). To most people, choosing a date for an appointment is not a very time-consuming task; one will consider it briefly, and if there is no apparent reason to favor one date over the other, one will realize that the decision is not terribly important and will choose in some random way. Damasio points out that this patient, unlike a person with functional emotions, could neither realize the triviality of the decision nor feel embarrassment at the annoyance he was causing the researcher by taking so long.
This example is just one of many instances in which emotion would have been a useful problem-solving tool. In his somatic marker hypothesis, Damasio suggests that in a normal person, there is a gut reaction (an emotion) that kicks in before a person starts to do any formal reasoning about the problem at hand. This gut reaction tells the person what kind of solutions to immediately avoid and which to pay more attention to, so that she does not have to go over every single possibility. In essence, this feeling narrows down the options from which the person needs to choose. Manfred Spitzer seems to agree with this hypothesis. He says that "the value of [an] object is judged very quickly, long before...a complete, detailed, full-color percept of the object has been worked out by the respective cortical area" (Spitzer 1999, p. 301). Additionally, according to Damasio, emotion achieves not only a narrowing-down of options, but it also boosts the attention and short-term memory necessary to solving problems once the person has moved on to reasoning after the original gut reaction. These things are necessary in order to keep track of all of the options and consequences that one has taken so much energy to figure out in the first place, so that a good decision can eventually be reached.
People with frontal lobe damage, and consequently impaired emotionality, cannot use the somatic markers suggested in Damasio’s hypothesis. Damasio (1994) gives a compelling example of the impact this has on ventromedial frontal lobe patients. He describes an experiment in which two sets of people -- frontal lobe patients and a normal control group -- were put in a gambling situation where they picked cards from any of four piles and received either rewards (fake money) or penalties (fines). Although the subjects were not told so in the beginning, two piles of cards were "high risk" -- many of the cards gave high rewards but some gave harsh punishments -- and the other two were lower risk -- the cards gave smaller rewards and smaller punishments. As the game went on, normal subjects chose the low-risk piles progressively more often and did relatively well, while the frontal lobe patients chose the high-risk piles progressively more often and ended up severely in debt. Damasio attributes this result to a "myopia for the future," which he believes stems from both an inability to keep in working memory the future outcomes to current actions, and a lack of the gut reactions that grew stronger and stronger over the course of the game for the normal subjects.
While the activity in this example is just a game, its implications are far-reaching. Damasio points out that this game is like many decisions in the real world: "It is full of uncertainty, and the only way to minimize that uncertainty is to generate hunches, estimates of probability, by whatever means possible" (Damasio 1994, p. 215). In fact, one can see the same sort of self-destructive behavior exhibited in this example in the everyday actions of both Phineas Gage and Damasio's patient, Elliot. After the damage to their frontal lobes, both men became unmotivated, easily distracted, and unable to plan for the future. These characteristics can be traced to the reasons previously described -- they no longer had attention-boosting somatic markers telling them which things were more important to do, so they were unable to do one thing and finish it. They were not able to build up long-term hunches about which plans of action would be good or bad, and so they never really chose anything. They simply did whatever was immediately interesting.
Values: they're not just for biological beings anymore!
Even in the field of robotics, values have shown to be useful. In her robots, Patti Maes (Broude 2003) implements "motivations": curiosity, laziness, hunger, thirst, aggression, fear, and desire for safety. Each of these motivational states can affect other behaviors. If a robot is hungry, for example, it will attempt to find food. In this way, motivation acts as an arbitration system in the robot, much in the same way as emotion seems to do in humans. Even in other types of robots, a set of values can be useful. In a Brooks-style subsumption architecture, for instance, there is a hierarchy of behaviors; in essence, behaviors are listed in order of importance, or value. Without some sort of system to tell it what to favor, what to avoid, and how to decide between different possible options, the robot would never decide to do anything at all.
The evolutionary standpoint
Beyond all the evidence that shows why values are so useful and what happens without them, there also exists the simple fact that a system of emotions has evolved in humans as well as other biological beings with perception-cognition-action systems; and if it has evolved, there must be some use for it. Evolution takes such a long time that, had emotion not done something that helped its possessor to survive, it would have just died out. One can see its usefulness in numerous situations involving instinct. For instance, if a person is suddenly approached by a bear in the woods, he or she will probably have an immediate feeling of fear. This is instantiated by the fight or flight response, which includes increased blood flow and breathing rate. This feeling allows the person to quickly decide to either fight the bear or run away. (cf. Mer's writeup for another example.) If there were no such emotional mechanism in place, the person would have to think through different possible options; by the time he had thought them all through, the bear would have killed him. It is notable that in this case, a person with frontal lobe damage probably would run away. According to Damasio's experiments, frontal lobe patients do have emotional reactions to the current situation -- they gave normal skin conductance results when startled, and in the case of the gambling game, they chose the piles that might give the highest immediate rewards. It is a build-up of emotion, one that would help them make plans for the future, that they lack.
What does it all mean?
There is no doubt that perception-cognition-action systems function with values. Values, in the form of emotion, allow people to select and perform tasks according to importance, make decisions based on future consequences, and react quickly in dangerous situations. Although emotion can sometimes get in the way of reason, people (and robots, and other perception-action-cognition systems) would be virtually paralyzed without it -- without having a feel for what is most important, one would spend ridiculous amounts of time deciding between tasks that might not even have any real impact. Patients with frontal lobe damage, such as Phineas Gage and Elliot, have managed to survive after their emotions have been wiped out; but one is inclined to think that they may not have been so lucky if they had not had people to take care of them. Even with people taking care of them, they functioned far from normally; they were so unmotivated that they could not keep steady jobs. Furthermore, if they had had absolutely no value system (instead of the short-term value system they did have), they may not have lived long. It would not matter to them that, for instance, a hot stove would burn them; and they would be extremely prone to injury. So, it would be very difficult, if not completely impossible, for a perception-cognition-action system to function without any values at all.
Broude, Gwen (2003). Handout from class: "Patti Maes on Emergent Functionality".
Broude, Gwen (2003a). Class lecture, March 6.
Damasio, Antonio (1994). Descartes' Error. New York, New York: Harper Collins Publishers.
Plotkin, Henry (1993). Darwin Machines. Cambridge, Massachusetts: Harvard University Press.
Pfeifer, Rolf and Christian Scheier (1999). Understanding Intelligence. Cambridge, Massachusetts: MIT Press.
Spitzer, Manfred (1999). The Mind Within the Net. Cambridge, Massachusetts: MIT Press.