Vanguard Project - Part 1: A Neuroeconomic Theory

The Mind’s Eye | 1

Running Head: The Mind’s Eye

The Mind’s Eye Chris Frueh Rutgers University

The Mind’s Eye | 2 Abstract The Purpose of this paper is to trace the process of situation-pattern-recognition through the various function-modules of the brain in order to define a subconscious neural framework for the problem-solving-by-analogy theory. I will present an experimental procedure to highlight, via functional imagery, the subconscious pattern recognition and resolution response to the brief exposure of a complex problem state. In the second part of the paper, I will elaborate a framework to explain the possible results in light of an ‘expected utility’ function married to a computer-housed virtual recognition protocol. This, I will fold this system into a model for a goal-oriented, intuitionreplicating artificial intelligence. To conclude, I will highlight future research possibilities to further confirm or concretely refute the framework foundational to this paper, that of the ‘module’ theory of brain activity.

The Mind’s Eye | 3 Introduction The branch of problem solving confronted herein is loosely defined as ‘complex problem solving by analogy’ requiring goal-oriented decision-making based on the brief exposure to the subject of the situational characteristics and requiring a rapid response informed by recollection of previously encountered scenarios of some similarity. Problem Solving Through Hypotheticals

(Politser 14) The above illustration details the traditional view of how a problem state’s resolution is attained, namely through the mental projection of hypothetical outcomes and the analysis thereof. I, however, would like to tweak this model somewhat as it is far too slow (relying on System 2 type of rational thought) and is vague on the influence of memory. Instead, I would postulate that the node marked ‘Perception’ is the priming effect that readies the access to similar scenarios (Kahnemann, 2003, p 452-3), that the entire ‘Hypothetical’ loop terminating in the actual decision is encapsulated in the subconscious memory of previously successful action-outcome pairs and that the successful analogy from the current to the remembered is a brief System 2 intrusion (at

The Mind’s Eye | 4 the Diagnosis node) into an exclusively System 1 task. This model would account for the rapid-response type of intuition exhibited during a mutating ambiguous problem state over an unbounded spatial area. The second area that requires introduction is the field of neuroeconomics. (Politser, 2007, p. 32) Put briefly, neuroeconomics is a neuroscientist’s examination of the brain during economic activities. The particular application to this problem is an attempt to rework the model of analogy-driven problem solving through the use of neural imaging and optimization (similar to Bayesian inference). Within the experiment below, the skeletal form of an expected utility function is the outline I assume, for the sake of argument, the subject follows.

Experimental Procedure The structure of the experiment will be the focused on the neurological response to a brief exposure of a complex problem state during the subject’s formulation of a response to the problem. In particular, I will flash an image of a chessboard with the pieces set up in a variation of the Sicilian Defense (or the Winawer variation of the French defense) known as the Poisoned Pawn. The reason this particular problem was chosen was because of the possibility of a favorable solution for either an aggressive1 or a defensive response.2 The subject(s) will be one or more experienced chess players. This condition is necessary to prevent the pattern recognition function to be engaged by a novice player recognizing the pieces themselves instead of the pattern of placement.

1 2

http://www.chessgames.com/perl/chessgame?gid=1044728 http://www.chessgames.com/perl/chessgame?gid=1044724

The Mind’s Eye | 5 The Poisoned Pawn Variation of the Sicilian Defense

This experiment will take two parts. After establishing a baseline for the subject’s brain at rest, I will expose him or her to one of the two variations of the gambit while he is attached to an fMRI. The readings will begin upon exposure of the stimulus and will end when he or she verbalizes a response to the situation (i.e. aggress or retreat) This will, in effect, calibrate the later readings to the particular layout of the subject’s brain patterns but without the time-sensitive readings that allow for a correlative hypothesis about the path of the information. I will record the areas that showed above average blood flow and, under the assumption that these areas played a role in the pattern recognition, will focus a second repetition using more time-sensitive equipment in order to support this assumption and the overall hypothesis.

The Mind’s Eye | 6 In the second part, I will repeat the previous protocols but with the subject attached to an EEG machine instead of an fMRI and using the gambit variation that was not used in the first part in order to keep the recognition based on playing experience rather than short-term memory. Again, the recording will begin at exposure and end at verbalization but, in interpreting the neural activity results, the response will be presumed to have been given upon activation of the speech centers of the brain. This, I believe, would provide a more concise time frame for the exercise. The recorded results will be, simply, where activation occurred during both exposures and when it occurred as measured during the second exposure. I expect initial firings to be in the occipital cortex as the subject views the situation. From there, I expect information to be sent to the parietal cortex as the various visual-spatial coordinates of the pieces are catalogued. Next, I expect firings to occur in the temporal lobe as the similarity of the situation acts like a priming stimulus to bring ‘to mind’ the previously encountered scenarios from the subject’s experience. This should be rapid and subconscious, fitting with Kahnemann’s (2003) placement of similarity firmly within the bounds of System 1 thought processes. (p. 453) As the situation is recalled, I expect activity in the anterior cingulate cortex as goals are analyzed in relation to previous problem states. Simultaneously, there may also be activity in the various portions of the brain that evaluate hypothetical outcomes (comparing the hypothetical outcomes with previously encountered outcomes) such as the ventromedial prefrontal cortex which has strong ties to predictive utility. This pattern comparison is also subconscious as postulated by Rosenblatt and Thickstun. (Rosenblatt, 1993, pg. 700) As this problem solving function begins to move towards production of a response, I expect significant activity in

The Mind’s Eye | 7 the orbitofrontal cortex or the amygdala as the subject weighs the salience of uncertainty or ambiguity. While the effect of risk aversion will not alter the hypothesis, it is important to note the role of the neural structures that house the risk-aversion functions and the time-specific point of their impact on the decision-making process. Lastly, I expect the firings to converge on the frontal cortex’s executive decision center before a response is vocalized.

Conclusion While the primary benefit of this experiment would be confirmation or denial of the theory of brain function localization, a significant secondary benefit would be a framework on which cognition-mimicking software might be based. For example, pattern comparison and organization is a task computers have been handling since the turn of the century, but, if this framework is valid, AI programmers might adapt these sorts of functions for use in a pattern-recognition-based machine capable of learning from previous scenarios (as the visual-spatial coordinates that the parietal cortex analyzes are written into the memory as a new situation able to be remembered). The priming effect would be simulated by a pattern comparison formula and the analogy-inferring process might be replicated by an organization or optimization formula. This, however, is contingent upon the modular theory of brain functionality and, more inhibiting, a coding structure capable of self-referential or self-modifying coding. Further research that might be done to examine the possibility of the theories hypothesized herein might be done by dissecting brain matter. Following in the footsteps of Wernicke, if the recognition function is localized to these specific brain structures,

The Mind’s Eye | 8 there will probably be bundles of neurons between these areas similar to the ‘path’ between Wernicke’s and Broca’s areas. (Cognition, Brain, and Consciousness, 2007, pg, 6) This could expand the ability of neuropathologists to understand brain damage: if damage is not localized to a specific structure but, rather, is between two structures, one might draw conclusions about the role of the previously connected structures from the symptoms in evidence after the damage.

The Mind’s Eye | 9 Works Cited Chen, S. (2008, November). Software-Agent Designs in Economics: An Interdisciplinary Framework. IEEE COMPUTATIONAL INTELLIGENCE MAGAZINE, 18-22. Foxall, G. R. (2008). Reward, emotion and consumer choice: from neuroeconomics to neurophilosophy. Journal of Consumer Behaviour, 7, 368-396. Gill, T. G. (2008). A Psychologically Plausible Goal-Based Utility Function. Informing Science: the International Journal of an Emerging Transdiscipline, 11, 227-252. Kahneman, D. (2003). A perspective on judgment and choice: mapping bounded rationality.. Am Psychol, 58(9), 697-720. Politser, P. (2007). Neurorationality: A Neuroeconomic Approach to Mental Health and Good Sense. New York: Oxford University Press, USA.

Rosenblatt, A., & Thickstun, T. (1993). Intuition and Consciousness. Psychoanalytic Quarterly, 63, 696-714. Rustichini, A. (2008). Dual or unitary system? Two alternative models of decision making. Cognitive, Affective, & Behavioral Neuroscience, 8(4), 355-362. (2007). Cognition, Brain, and Consciousness: Introduction to Cognitive Neuroscience. Toronto: Academic Press. (1995). Complex Problem Solving: The European Perspective. Mahwah, NJ: Lawrence Erlbaum.