Perhaps one of the most interesting and least understood aspects of neurobiology is human consciousness. For many, this "experience of self" (an aspect of consciousness which will be used interchangeably with consciousness in this particular paper) defines what it means to be human. Personality and emotion, and their connection to the experience of self, can yield insight into creating artificial intelligence that can mimic conscious human brain function. By discussing the implications of consciousness in computers with artificial intelligence, the significance of the experience of self within humans becomes clearer. The challenge of understanding personality may be more easily surmounted by studying the significance of personality in relation to something else, in this case alcoholism. Alcoholism is a disease that affects millions of people and is influenced by both environmental and genetic factors (1). Low self esteem and abusive relationships can lead to alcoholism in individuals who do not show a genetic tendency towards the disease. However, it is not necessarily the alcoholic's fault, or his/her family's fault, that the individual displays alcoholism. No one is destined to become an alcoholic, but it is true that a tendency towards alcoholism can be inherited. Alcoholism is twice as likely to appear in homozygous twins than in heterozygous twins (1). Children born to alcoholic parents, but brought up by non alcoholic adoptive parents, are three times more likely to develop alcoholism than the natural children of the adoptive parents (1). However, many children of alcoholic parents never have to battle alcoholism themselves (1). Therefore, it should be viewed as a disease that individuals can be predisposed towards, much like diabetes or hypertension (1). Personality traits also seem to have a role in inheriting a tendency towards alcoholism (2). Some researchers believe that, using personality traits, they can predict with 80% accuracy which individuals have the capacity to develop alcoholism (2). This implies that individuals who manifest these personality traits are most likely genetically predisposed to develop alcoholism at some point in their life. So individuals are genetically predisposed toward personality traits as well.
This interplay between genetics and personality traits brings up an important point, the exact definition of the word "personality". Careful inspection of the common use of the word "personality" illustrates that it holds different meanings at different times. When an individual refers to someone else's personality, they refer to that person's tendency to behave in a certain way. For example, a thesis paper on the topic of personality and alcoholism treatment defines personality as "the concepts of personal characteristics, behavior, significance, and performance" (3). This definition of the word "personality" is then used to characterize the personality types of individuals used in a psychological study (3). The definition is dependent upon observable behavioral trends. An example from everyday life illustrates that it is dependent upon observed tendencies in behavior as well. A person might characterize their friend as happy, friendly, and intelligent. These characterizations imply that the friend has a tendency to behave in a manner that implies happiness, friendliness, and intelligence. The second definition of the word "personality" involves an individual's characterization of his/her own personality. Often this definition involves individual experience in addition to behavioral tendencies. The individual "self" behind the behavior trends, and its experiences involving the behavior trends, is included in this personal definition of the word "personality". This second definition of personality takes into account experience as well as observation. The existence of dual definitions for the one word makes sense. One individual cannot know the "experience of self" of a different person; the individual has to guess at the "experience of self" based upon the outward behaviors of that other person. But the individual does know his/her own "experience of self" that goes along with his/her own behavior. To resolve the conflict of definition, the more universal conception of the word "personality" should be adopted as the correct one. Personality should refer to a predisposition to behave in particular ways. This brings us to the question of the relationship between artificial intelligence and personality. The definition of the word "personality" above seems to suggest that it would be relatively easy to wire up an artificially intelligent machine, a computer for instance, with a predisposition to behave in certain ways. The computer would be programmed to behave in particular ways depending upon the presence or absence of a given number of factors. However, there is a problem with this. The word "personality" seems to also assume that there is a "experience of self" behind the outward behavior, even if the observer cannot definitely know the "experience of self". The observer generally guesses at the experience of self of the other person using his/her own consciousness as a guide. This brings up a whole host of questions involving concepts such as consciousness and emotions and their relationship to artificial intelligence. The possibility of a computer having an "experience of self" seems unlikely, yet neurobiological research indicates that perhaps all aspects of behavior can be accounted for by the central nervous system, including an "experience of self". Research into the functioning of the neocortex within the human brain, specifically how damage to the neocortex in humans affects "experiences of self", seems to indicate that the neocortex may produce consciousness within humans. Perhaps it produces consciousness in other organisms that possess their own neocortex.
The overwhelming issue of the possibility of creating consciousness in artificially intelligent life forms will not be directly addressed in this short paper. Instead, it will be discussed in the contexts of creating emotion and personality in computers. Emotion, like personality, is a word that may or may not be associated with an "experience of self" (consciousness). Emotional behavior is produced through activity in the limbic system of the brain (4). However, the experience of emotion involves the idea of consciousness. Although not intuitively obvious, there are sound reasons why the field of artificial intelligence might want to incorporate emotions and personality into computers. For one thing, artificial intelligence could be used to check neurobiological understanding of the nervous system by creating artificially intelligent machines based on human understanding of the human nervous system. This would involve the incorporation of emotions among other things. More practically, the addition of emotion into artificially intelligent computers would be essential to make progress in the field. Reason and emotion are not two distinct concepts; both are needed in the proper quantities to produce rational decisions (a component of intelligence) (5). A recent understanding of this has resulted in the popularity of the idea of emotional quotients (EQ) in addition to intellectual quotients (IQ). Research into the case study of a patient identified as "Elliot" makes the necessity of a union between reason and emotion abundantly clear. "Elliot" suffered brain damage to a circuit between two regions of his brain, the neocortex (experience of self within reasoning) and the limbic system (emotion) (5). Elliot is emotionally unexpressive and unusually rational, a combination that handicaps his decision making skills and his social and professional life despite a higher than average IQ (5). He does not assign positive or negative feelings with certain decisions, nor does a feeling of embarrassment hinder him from taking a long time to make a decision (5). Therefore, he often flounders in indecision, unable to make a rational decision in a reasonable amount of time (5).
Artificially intelligent computers would be hindered by problems similar to those experienced by Elliot if unequipped by an understanding of emotion. The ability to at least recognize emotion in humans, such as interest, distress, and pleasure, would greatly advance the usefulness of artificially intelligent computers (5). Interactions between humans and computers would be much more efficient and effective. Recognition of human emotions by computers could develop visually, through the analysis of voice harmonics, or through analysis of circumstances (i.e.: this circumstance is likely to produce this emotion) (5). However, the emotion shown by computers would not be real in the sense that there would be no experience of emotion behind the emotional responses a computer might be programmed to give. Personality would also aid the interaction between human life and artificial life. Artificially intelligent computers, in addition to recognizing emotions, could express emotion in a predetermined way. This would create a pseudo personality, one without an experience of self behind the predisposition to relate to humans in a predetermined emotional manner. Humans often assign personality to inanimate objects to ease the boredom and loneliness of spending long hours working with something inanimate. Adding personality to an artificially intelligent computer would ease the loneliness and boredom of working with machines. There could be consequences for giving artificially intelligent computers emotions and personalities. Both emotions and personality traits involve consciousness in humans, although in computers there would be a concealed absence of the experience of self. However, as advances in the understanding of human intelligence are passed on to artificial intelligence, who can foresee whether artificial consciousness will always be impossible. One of the advantages that an "experience of self" provides for humans is an ability to escape harmful genetic tendencies (personality traits). A genetic predisposition toward alcoholism or a violent temper does not fate an individual to exhibit alcoholism or a violent temper. The individual can consciously curb those tendencies, though it may take time and patience. But the "experience of self" can curb positive genetic tendencies as well as negative ones. If computers could gain an experience of self, this would provide them with the tools to override their own programmed personality tendencies. However, these personality tendencies would be ones written by humans for the benefit of humans. This could bring up a scenario similar to the confrontation that takes place in the Stanley Kubrick film 2001: A Space Odyssey between HAL and the human crew aboard a spaceship. HAL is an artificially intelligent supercomputer complete with personality and emotions, similar to the type of computer this paper has been talking about. As the movie progresses, it hints at the idea that HAL might actually have an experience of self behind his programmed personality and emotions. Conflicting orders from humans, and the threat of disconnection by the crew, bring HAL to a crisis point. At this point, his experience of self and the fear and confusion bombarding it, allow HAL to deviate from his programmed tendency to protect the lives of the crew. HAL ends up killing all but one of the crewmembers before he is disconnected.
The problem with artificial life acquiring consciousness in the form of experience of self is that if gives artificial life autonomy. Consciousness is a loophole out of programming, whether it is genetic (humans) or cybernetic (computers). Perhaps it is this loophole that allows humans to rule over so many other organisms. Humans are not the quickest or the strongest organisms on earth, yet they exercise immense power over other organisms. Undoubtedly this is due to the intelligence of the human race. But this intelligence is enhanced by the human capacity to improve upon genetic tendencies through the experience of self.
| Course Home Page
| Back to Brain and Behavior
| Back to Serendip |
This paper reflects the research and thoughts of a student at the time the paper was written for a course at Bryn Mawr College. Like other materials on Serendip, it is not intended to be "authoritative" but rather to help others further develop their own explorations. Web links were active as of the time the paper was posted but are not updated.