32
Neuroscience as the Foundation of Game Theory Kaushik Majumdar Systems Science and Informatics Unit, Indian Statistical Institute, Bangalore Center https://sites.google.com/site/isi cng/ (Computational Neuroscience Group Home)

Neuroscience as the Foundation of Game Theory

Embed Size (px)

DESCRIPTION

Neuroscience as the Foundation of Game Theory. Kaushik Majumdar Systems Science and Informatics Unit, Indian Statistical Institute, Bangalore Center https://sites.google.com/site/isicng/ (Computational Neuroscience Group Home). Games and Neuroscience. Games played by players. - PowerPoint PPT Presentation

Citation preview

Page 1: Neuroscience as the Foundation of Game Theory

Neuroscience as the Foundation of Game Theory

Kaushik Majumdar

Systems Science and Informatics Unit, Indian Statistical Institute, Bangalore Center

https://sites.google.com/site/isicng/

(Computational Neuroscience Group Home)

Page 2: Neuroscience as the Foundation of Game Theory

Games and Neuroscience

• Games played by players.

• Strategies of the players are modulated by their behaviors.

• Behaviors are modulated by neurobiological build up of the players.

• So Neuroscience modulates playing strategies in the Games.

Page 3: Neuroscience as the Foundation of Game Theory

Subjective Desirability

U1(x) is the first player’s utility function in the two player inequality aversion game.

U1(x) = x1 – αID – βIA, where

X = [x1 x2], ID = max{x1 – x2, 0}, IA = max{x2 – x1, 0}, 0 ≤ β ≤ α < 1. α and β are sensitivities to disadvantageous and advantageous inequalities respectively.

Game theory and neural basis of social decision making, Lee, Nature Neuroscience, 11(4): 404 - 409, 2006.

Page 4: Neuroscience as the Foundation of Game Theory

Prisoners’ Dilemma with Inequality Aversion

Page 5: Neuroscience as the Foundation of Game Theory

Brain Areas Involved in Moral Sensitivity

Neural correlate of moral sensitivity, Moll et al., J. Neurosci., 22(7): 2730 – 2736, 2002.

Page 6: Neuroscience as the Foundation of Game Theory

Moral Emotion

• Moral emotions differ from basic emotions in that they are intrinsically linked to the interests or welfare either of society as a whole or of persons other than the self. Moral emotions are readily evoked by the perception of moral violations. It has been suggested that, in contrast to laborious deductive

Page 7: Neuroscience as the Foundation of Game Theory

Moral Emotion (cont)

reasoning, they enable rapid, automatic, and unconscious cognitive appraisals of interpersonal events.

More can be seen in:

• A. R. Damasio, Descartes' error: emotion, reason, and the human brain. Avon, New York, 1994.

Page 8: Neuroscience as the Foundation of Game Theory

Morality from a Different Angle

• Trading in nonhuman animals is ‘permitted’ under wide ranging conditions with the ambit of laws made by humans. Confining, slaughtering and subjecting them to various experiments are ‘legal’ in the human society under wide ranging conditions. But the scope of the same acts on the humans is much more restricted under the human legal systems.

Page 9: Neuroscience as the Foundation of Game Theory

Morality (cont)

• From an objective point of view, devoid of human selfishness, all these acts on human and nonhuman animals are equally justifiable or equally unjustifiable. Still the human laws prevail because humans have become hugely more powerful than all other species. That is why eating beef is moral but eating human meat is a criminal offence even for a tiger.

Page 10: Neuroscience as the Foundation of Game Theory

Power

• So legality or morality is decided by power. In a capitalistic society, where capital is the most powerful entity, gender or racial inequality may be immoral or unethical, but inequality between workers and capitalists is quite natural and acceptable.

Page 11: Neuroscience as the Foundation of Game Theory

Class vs. Power

• Power is a more objective entity than class. One class dominates the other simply because the dominating class is more powerful than the dominated.

• Analysis by power gives a finer resolution than analysis by class.

Page 12: Neuroscience as the Foundation of Game Theory

Power in Game Theory

• A player plays a game because he/she has the power to participate in the game with own strategies.

• A player who is substantially more powerful than the others, will destroy the independence of the players by dominating the other players.

Page 13: Neuroscience as the Foundation of Game Theory

Quantitative Definition of Power

• Here our aim is to define power for any general system.

• We will first define system momentum.

• Then we will measure the rate of change in the system momentum.

Page 14: Neuroscience as the Foundation of Game Theory

System Momentum

• Let a system be defined by a totally differentiable vector field f : Rn+1 → Rm. Then the system momentum is (∂Mf/∂t), where M Rm is a fixed weight vector.

• System power is defined by (∂2(Mf)/∂t2)(∂f/∂t), which is a straight forward extension of the notion of power in classical mechanics, where f is displacement and M is mass.

Page 15: Neuroscience as the Foundation of Game Theory

Revolution and Evolution

• When ∂f/∂t is large, i.e., a big change in a small time – called a revolution and when ∂f/∂t is small, i.e., a small change in a big time – called an evolution. Clearly a lot more power is needed to make a revolution happen compared to an evolution.

Page 16: Neuroscience as the Foundation of Game Theory

Power of Trend in Society

• In a social system f = (f1,……..,fm)T, where each component fi : Rn+1 → R signifies a trend, called the ith trend. M = (m1,……,mm), where each mi is the number of individuals associated with the trend. If all but the ith trend in f are fixed then (∂2(Mf)/∂t2)(∂f/∂t) = mi(∂2fi/∂t2)(∂fi/∂t), mi is the number of individuals following the trend.

Page 17: Neuroscience as the Foundation of Game Theory

Power of Trend (cont)

• If mi is large the power associated with the trend will be high.

• How to have a large mi then?

Simple, create a trend and gather people in favor of it!

Page 18: Neuroscience as the Foundation of Game Theory

Brain Activation During Lipreading

Buccino et al., J. Cognitive Neurosci., 16: 1 – 14, 2004.

Page 19: Neuroscience as the Foundation of Game Theory

Human Mirror Neuron System

Page 20: Neuroscience as the Foundation of Game Theory

Human MNS (cont)

• It has been hypothesized that the fundamental mechanism at the basis of the experiential understanding of others’ actions is the activation of the mirror neuron system (Gallese et al., 2004).

• Recently mirror neurons have been located in different parts of the human brain (Mukamel et al., 2010).

Page 21: Neuroscience as the Foundation of Game Theory

Dopamine Neurons

• Encodes goal-directed behavior (pathological case Parkinson’s disease).

• Activate reward expectation.

• Seek novelty.

Page 22: Neuroscience as the Foundation of Game Theory

Novelty & Reward Expectancy Hypothesis

x = novelty

iy = reward expectancy

Complex Plane

Page 23: Neuroscience as the Foundation of Game Theory

fn as a Discrete Dynamical System

fn : С → С, where n is the time.

P = M|{(fn + 1 – fn) – (fn – fn – 1)}(fn – fn – 1)| is the expression for the power. | | denotes modulus.

When P is high enough M is also increased.

Page 24: Neuroscience as the Foundation of Game Theory

Social Trend

Sequence {fn}n constitutes a Riemann Surface of individual trend.

At a particular time n the envelope of fn across all the individuals in the society will give the social trend.

Page 25: Neuroscience as the Foundation of Game Theory

Power Concentrates

010

2030

4050

0

20

40

600

0.5

1

1.5

2

2.5

Social noveltySocial expectation

Pow

er

010

2030

4050

0

20

40

600

0.5

1

1.5

2

2.5

Social noveltySocial expectation

Pow

er

Page 26: Neuroscience as the Foundation of Game Theory

Consequence

• In a society where power is more or less uniformly distributed among individuals and institutions, because of biological diversity among individuals, some are more likely to have stronger dopamine drive for novelty and reward expectation than most others. Eventually power will concentrate around them and the distribution will become unequal.

Page 27: Neuroscience as the Foundation of Game Theory

Multiple Centers of Power Concentration

010

2030

4050

0

20

40

600

0.5

1

1.5

2

2.5

Page 28: Neuroscience as the Foundation of Game Theory

Consequence

• Multiple powers will try to exert their influence on the same resources and a power politics or power game will ensue, where each one will try to dominate the other and over iterations of the game the independence of the players is likely to be compromised. Many realistic games must model this phenomenon.

Page 29: Neuroscience as the Foundation of Game Theory

Additional References

• C. Camerer, G. Loewenstein and D. Prelec, Neuroeconomics: how neuroscience can inform economics, J. Economic Literature, 43(1): 9 – 64, 2005.

• E. Fehr and C. F. Camerer, Social neuroeconomics: the neural circuitry of social preferences, Trends Cog. Sci., 11(10): 419 – 427, 2007.

Page 30: Neuroscience as the Foundation of Game Theory

References (cont)• http://en.wikipedia.org/wiki/Power_(philoso

phy)

• Galles et al., A unifying view of the basis of social cognition, TINS, 8(9): 396 – 203, 2004.

• M. Iacoboni and M. Dapretto, The mirror neuron system and consequences of its dysfunction, Nature Reviews Neuroscience, 7: 941 – 951, 2006.

Page 31: Neuroscience as the Foundation of Game Theory

References (cont)

• Mukamel et al., Single neuron responses in humans during execution and observation of actions, Curr. Biol., 20(8): 750 – 756, 2010.

Page 32: Neuroscience as the Foundation of Game Theory

Thank You