Extending HAL's basic system of Learning to include emotional development, perceptions of the self, and awareness of the self.
As mentioned at the beginning of this series, this is essentially a tabula rasa model. My approach is currently only to add preprogramming when necessary. This thesis takes the balanced view, such as suggested by Dupre (2002). I tried, but could not construct a tabula rasa model without introducing a basic set of four emotions. These are sufficient, as I cast them, for the computer to have self motivation to discover itself, and to learn about itself though play, as will be discussed in the next topic. However, first it is needed to provide an explanation of how the software creates new symbols such that it can invent its own self awareness, and the first part of that is describing how new symbols for emotions are created from the base set.
The first topic in this series, "Designing HAL (1): Hypercognition, Freud and Jung" integrated Demetriou's theory of hypercognition with theories of the unconscious from Freud and Jung. The next topic, "Designing HAL (2): A Learning AI with Emotions and Dreams," outlined a framework for designing a complete computer model of cognition, including emotions and dreams.
The academic issue which reviewers expressed for the system was first and foremost whether a computer model could effectively provide a simulacrum of the conscious/unconscious division. This topic therefore describes a way of implementing the development of unconsciousness in terms that would be applicable to software, although the paradigm of the model itself may still provide an answer to philosophical questions, such as the nature of reality and language.
In this model, there is no hard delineation between that which can be consciously known of the cognition process, and that which is buried in the deeper unconscious. The amount of conscious access to deeper symbols can be increased with greater effort. The amount of effort is defined by the control stage in the hypercognition model.
The functionality is as follows. Each symbolic object contains a scoreboard of values indicating its properties, one of which is 'depth.' Another value in the scoreboard is 'successfulness.' If a particular combination of symbolic states is successful, points are added to the 'successfulness' value. When the points exceed a success threshold in more than two objects for a particular cognition, a new symbolic object is created with links to the successful objects. Scoreboard points from the previously successful objects are then transferred to the new object. The new object is then assigned the same depth as the successful objects, and the depth of the successful objects creating the new object is pushed down.
The first actions are set by the six primal emotions defined in the previous section. Of these, the 'fear/anger', or 'fight/flight' responses, are the most primary. They are triggered most frequently by control, so they create the most new symbols. The illustration shows how the new object is created and connected.
The properties and methods of the basic symbolic object, underlying this process, is applicable to all six processing domains in the hypercognition model. It may have a direct neural correspondence, as mentioned before, but in physiology there may not be a direct correspondence between one 'symbol' and one neural cell. It may be possible to define a model where there is a correspondence between one symbol and a physical group of cells, but as there are 20 billion cells in the neocortex alone, it would be very difficult to do so.
In the hypercognition model, the symbols also do not correspond directly to words in any language. This is because linguistic processing domain develops later. By that time, there are already many new symbols derived from the original core symbols by new cognitions. Therefore, this model does not support the historical 'Platonic idea of form' as objects having some ideal nature as "things-in-themselves." Instead, words are attached by the act of 'dubbing' to a collection of properties, in accordance with the theories of Kripke (1980). Kripke's theories are themselves based on those of Russell (1905).
As the words 'fear' and 'flight,' for example, are assigned to large groups of symbols by the time language is learned, there is no direct correspondence between the word 'fear,' or 'flight,' and the original core symbol which led to the creation of the symbol collection to which the word is later assigned. As the sequence of experiences lead to the development of different networks of interconnected symbols in each person, there is no absolute way to define the words in relation to some ideal.
The reason, by this model, that the Platonic ideal of form is so attractive intuitively (in there being some real abstraction, such as 'justice,' to which all people have intuitive access, explaining the meaning of language) would appear to be that it is because of this development of new symbols from some core set. Therefore, it is appealing to think there is some 'core' concept to which all people have access. But the ideas of 'equality' or 'justice,' by this model, are developed experientially. The experiences are different for each person. Hence, the conclusions as to what some ideal concept of such ideas might be different for each person.
The Piagetian theory is based on the idea that we are born without the ability to differentiate consciously between our experience of ourselves and the external world. A baby has emotions, but the emotions drive autonomous, unconscious responses. With introspection, we develop a model of our own emotional responses. Our self perception initially augments, then with maturation of the persona in the perception of self, mostly overrides the autonomic response.
The dilemma presented to a newborn child is that there is initially no value system attached to the emotional states and the autonomous responses, so as consciousness of one's own control over how to respond to exchanges in the environment evolves, the child has to learn whether the decisions made are effective or ineffective.
There are thus multiple types of indirect connections in cognition between the stimulus and response, which has made psychological experimentation, to determine the internal nature of our comprehension and decision-making processes, extremely difficult.
Initially, at conception, there is no self perception. Intuitive emotions entirely control the executive function of cognition. With introspection, the mind is able to observe the state of the emotions after they cause an intuitive response. the higher-level objects in the intuitive emotions are more accessible, so the first self-perception creates one symbol, of like/dislike (abstracted as a continuous value on a single axis, rather than as separate symbols which can fire in parallel). Each act of introspection creates a new symbol in the self perception. Subsequent introspections may add more symbols closely associated with the like/dislike value, and add additional lower-level symbols.
After the formation of a like/dislike association, the instinctive responses no longer pass through the motivation center unaffected,but are moderated by the impulses received from self-perception.
If the results continue to be successful so much that its success value exceeds its control threshold, the new like-dislike symbol creates a new symbol, just as described earlier; but the new symbol becomes part of the intuitive complex, rather than remaining in self-perception. This is how the individual can train intuitive response to perform complex tasks, such as driving a car.
Newly created symbols in self-perception contain not only the intuitive responses, but other perceptions from the other processing domains. After an individual has created some values in self=perception, the individual can introspect on the self-perception itself. This creates a symbol representing one's own self-perception, which is the formation of the concept of self. Introspection on the concept of self is self awareness.
The process of self awareness is exactly the same as for self perception, except now the created symbols are not observations of one's own state in other respects, but rather, observations of one's own self perceptions.
The cognitions of one's self awareness have an additional significance in the hypercognition model. Just as self perceptions permit one to control one's motor responses consciously, self perceptions unconsciously control the congruence of input of other information from the processing domain, including emotions. If there is dissonance between other information from the processing domain and our self perception, we experience a conscious emotional conflict. Later self-concepts such as doubt, uncertainty, and anxiety are associated with this conflict, depending on the nature of the incongruence. Similarly, if our self perception is in congruence with other perceptions, we have positive associations, which we later associated with belief, certainty, and confidence, as our self-awareness increases.
If you are unaware of this long-standing objection, the 'introspection illusion' refers to a cognitive bias whereby people wrongly believe they know their own mental states, while treating others' introspections as unreliable.Sometimes people therefore make confident, but false, explanations of their own behavior, or inaccurate predictions of their future mental states (Nisbett, 1977). This illusion has led some to state that a computer model cannot truly be self aware, because introspection is unreliable.
Rather than being impossible, this model actually explains how the 'introspection illusion' is caused. This is because the executive function, in this model, indeed does *not* have direct insight into mental state. I cannot take credit for it, because it's simply a version of how introspection works from the historically well-regarded "Critique of Pure Reason" (Kant, 1787).
The 'perception of self' is separate from 'self awareness" and the 'point of consciousness.' The 'perception of self' is a simplified image of the self, created by introspection on one's state. The simplified image of the self includes only that much of the consciousness as accessible by the degree of effort in the introspection. The perception of self generates congruence or dissonance with the unconscious emotional state, but the perception of self is not directly accessible to the executive function, where the current point of consciousness resides in this model. Introspection on the perception of self does enable evaluation of the significance of the emotional state, but does not provide self awareness of one's own self.
When introspecting on the 'perception of self,' another derivative, simplified image is created, and *that* is the information of self awareness that the executive function can access.
The processing domains only define the information which 'can' be accessed by consciousness. The processing domains also include information which remains unconscious. the processing domains do not include the 'current point of consciousness.'
That is, in this model:
- The processing domains contain that which *can* be part of consciousness.
- The perception of self is the direct product of introspection
- The perception of self results in congruence/dissonance
- Congruence/dissonance affects motivation
- Self awareness is a secondary product, created by Introspecting specifically on the perception of self alone
- Self awareness is the only abstraction of the self which is accessible to the current point of consciousness .
The current point of consciousness is in the executive function, above the the processing domains. Unless the executive function has introspected on its self perception sufficiently that the self awareness, created by the introspection on self perception, has evolved to the point of knowing the self perception is a higher-level and simplified abstraction of lower symbols, then the executive function will produce the errors described as the 'introspection illusion' by Nisbett.
Later, this topic will be expanded with some a detailed description of software properties and methods for the creation of new concepts, that is, the base class of objects which will be shared across all instances in the processing domain. First, additional abstracts will be formulated on the specific additional capabilities in each of the processing domains, and then the characteristics of functions in the other two stages of cognition.
- Aristotle (ca. 360BC)On the Soul
- Demetriou, A. (2006)Neo-Plagetian Theories of Cognitive Development.
- Demetriou, A ; Spanoudis, G.; Mouyi, A. (2006) "A Three-level Model of the Developing Mind: Functional and Neuronal Substantiation"
- Dupre, J. (2002) "Making Hay with Straw Men" American Scientist
- Festinger, L. (1957).A Theory of Cognitive Dissonance.California: Stanford University Press.
- Freud, S. (1933) New Introductory Lectures on Psychoanalysis, Penguin Freud Library
- Freud, S. (1940) An Outline of Psycho-analysis
- Henderson, D.; Jacobson, S.; Johnson, A. (2002) "The Theory and Practice of Simulated Annealing."
- Jung, C. G. (1935)Structure & Dynamics of the Psyche,The Collected Works of C. G. Jung, Volume 8. Princeton University Press
- Jung, C.G. (1971).Psychological Types, Collected Works, Volume 6, Princeton, N.J.: Princeton University Press
- Kripke, S. (1980)Naming and Necessity. Harvard University Press:
- Locke, J. (ca. 1690)Essay Concerning Human Understanding, Volume 2
- Nisbett, R.E.; Wilson, T.D. (1977) "Telling more than we can know: Verbal reports on mental processes".
- Pinker, S. (2002)The Blank Slate: The Modern Denial of Human Nature. Continuum
- Plato (380 BC)The Republic, Allegory of the Cave, 514a–520a
Some have asked for some more references, I am still compiling it, but here is a start.
- Aristotle (ca. 360BC) On the Soul. http://classics.mit.edu/Aristotle/soul.html
- Asch conformity experiments: http://en.wikipedia.org/wiki/Asch_conformity_experiments
- Baddeley, A; Sala, S. D. (1996) "Working Memory and Executive Control," Philosophical Transactions: Biological Sciences 351.1346, Executive and Cognitive Functions of the Prefrontal Cortex http://www.jstor.org/stable/3069185
- Breazeal, C.; Gray, J.; Berlin, M. "An Embodied Cognition Approach to Mindreading Skills for Socially Intelligent Robots" http://ijr.sagepub.com
- Breazeal, C.; Thomaz, A. (2005)Learning from Human Teachers with Socially Guided Exploration http://www.cc.gatech.edu/~athomaz/papers/BreazealThomaz-ICRA08-final.pdf
- Chomsky, N. "Tool Module: Chomsky's Universal Grammar"
- Cowan,N. (2010) Multiple Concurrent Thoughts: The Meaning and Developmental Neuropsychology of Working Memory Dev Neuropsychol.
- Cannon, W.B. (1915) "Bodily Changes in Pain, Hunger, Fear and Rage: An Account of Recent Researches into the Function of Emotional Excitement." Appleton.
- Davidson, D.; Suppes, P.; Siegel, S. (1957) Decision-Making: An Experimental Approach. Stanford University Press.
- Davidson, D (2001). Essays on Actions and Events. Oxford University Press.
- Demetriou, A. (2006) Neo-Plagetian Theories of Cognitive Development. https://www.academia.edu/2090948/Neo-Piagetian_theories_of_cognitive_development
- Demetriou, A ; Spanoudis, G.; Mouyi, A. (2006) "A Three-level Model of the Developing Mind: Functional and Neuronal Substantiation" https://www.academia.edu/2065614/A_Three-level_Model_of_the_Developing_Mind_Functional_and_Neuronal_Substantiation
- Dupre, J. (2002) "Making Hay with Straw Men" American Scientist http://www.americanscientist.org/bookshelf/pub/making-hay-with-straw-men
- Festinger, L. (1957). A Theory of Cognitive Dissonance. California: Stanford University Press.
- Freud, S. (1933) New Introductory Lectures on Psychoanalysis, Penguin Freud Library
- Freud, S. (1940) An Outline of Psycho-analysis http://archive.org/stream/outlineofpsychoa027934mbp/outlineofpsychoa027934mbp_djvu.txt
- Haidt, J. "The Emotional Dog and its Rational Tail: A Social Intuitionist Approach to Moral Judgment," University of Virginia http://www.motherjones.com/files/emotional_dog_and_rational_tail.pdf
- Henderson, D.; Jacobson, S.; Johnson, A. (2002) "The Theory and Practice of Simulated Annealing." http://homes.ieu.edu.tr/~agokce/Courses/Chapter%208%20Theory%20and%20Practice%20of%20simulated%20Annealing.pdf
- Jung, C. G. (1935) Structure & Dynamics of the Psyche, The Collected Works of C. G. Jung, Volume 8. Princeton University Press
- Jung, C.G. (1971). Psychological Types, Collected Works, Volume 6, Princeton, N.J.: Princeton University Press
- Kant, I..(1787) Critique of Pure Reason.
- Kripke, S. (1980) Naming and Necessity. Harvard University Press:
- Kruger and Dunning "Unskilled and Unaware of It: How Difficulties in Recognizing One's Own Incompetence Lead to Inflated Self-Assessments," Journal of Personality and Social Psychology, 77 (6): 1121–34. doi:10.1037/0022-35220.127.116.111 http://en.wikipedia.org/wiki/Dunning%E2%80%93Kruger_effect
- Locke, J. (ca. 1690) Essay Concerning Human Understanding, Volume 2. http://www.gutenberg.org/ebooks/10616
- Maslow, A.H. (1943). A theory of human motivation. Psychological Review 50 (4) 370–96. Retrieved from http://psychclassics.yorku.ca/Maslow/motivation.htm
- Milgram experiment: http://en.wikipedia.org/wiki/Milgram_experiment
- Nagel, T. (1974) "What Is It Like to Be a Bat?" The Philosophical Review, Vol. 83, No. 4 (Oct., 1974), pp. 435-450
- Nisbett, R.E.; Wilson, T.D. (1977) "Telling more than we can know: Verbal reports on mental processes". http://deepblue.lib.umich.edu/bitstream/handle/2027.42/92167/TellingMoreThanWeCanKnow.pdf
- Nisbet, E.; Garrett G. (2008) "Belief in rumors Hard to Dispel: Fact checking easily undermined by images, unrelated facts," Ohio State University. http://www.comm.ohio-state.edu/kgarrett/FactcheckMosqueRumors.pdf
- Nuxoll, A. (2007) Enhancing Intelligent Agents with Episodic Memory http://deepblue.lib.umich.edu/bitstream/handle/2027.42/57720/anuxoll_1.pdf;jsessionid=17B337D6E41731B075E9C653EA0CF1A7?sequence=2
- Pinker, S. (2002) The Blank Slate: The Modern Denial of Human Nature. Continuum
- Plato (380 BC) The Republic, Allegory of the Cave, 514aâ€“520a
- Prasad, M., Perrin, A; Bezila, K; Hoffman, S.; Kindleberge, K; Manturuk K; Powers, A. (2009) "There Must Be a Reason”: Osama, Saddam, and Inferred Justification," Sociological Inquiry, 29.2, pages 142–162 http://www.sciencedaily.com/releases/2009/08/090821135020.htm
- Raaijmakers, J.G (1993) "The story of the two-store model of memory: past criticisms, current status, and future directions". Attention and performance. XIV (silver jubilee volume). Cambridge, MA: MIT Press
- Russell, Bertrand (1905) On Denoting. Mind.
- Sampson, G. (2005) Educating Eve: The 'Language Instinct' Debate. continuum.
- Stanford Prison Experiment: http://en.wikipedia.org/wiki/Stanford_prison_experiment
- Thorndike E.L.; Woodworth R. (1901) The Influence of Improvement in one Mental Function upon the Efficiency of Other Functions. Psychological Review, 8, 247-261.
- Weizenbaum, J. (1976) Computer Power and Human Reason: From Judgment To Calculation. San Francisco: W. H. Freeman