The Neural Basis of Decision-Making During the Ultimatum Game in Adults With

Schizophrenia

Keith Ziegler

Rochester Institute of Technology

Abstract

The Ultimatum Game is used to measure how feelings of fairness influence decision making. People do not generally accept incredibly unfair offers even if it would maximize their utility. People with schizophrenia accept unfair offers more often. Both schizophrenia and non-schizophrenic participants will play the Ultimatum Game in an fMRI machine against both human and computer opponent. It is predicted that neurotypicals will accept more unfair offers when playing against a computer. There will be less activation of the bilateral anterior insula in those with schizophrenia when compared with those without schizophrenia in all opponent conditions. Finally, schizophrenia participants will show no brain activation difference between opponent conditions. A study like this has not been performed with participants with schizophrenia.

The Neural Basis of Decision-Making during the Ultimatum Game in

Adults with Schizophrenia

        Economics and Game Theory try to explain decision making as a purely logical process where emotions play no part (Tadelis, 2013). However, humans sometimes base their decisions on spite or some other emotion. The merger of economics and psychology, called behavioral economics, examines decision making under various conditions, and why people make irrational decisions (Tadelis, 2013). Neuroeconomics is the study of which parts of the brain are activated or inhibited during decision making (Glimcher & Fehr, 2014). Damage to the brain through either trauma or a mental disorder such as schizophrenia can change how and where in the brain decisions are made (e.g., Sanfey, Rilling, Aronson, Nystrom & Cohen, 2003). Research by Larquet, Coricelli, Opolczynski, and Thibault (2010) suggests that those with positive schizophrenia symptoms cannot anticipate negative consequences of their decisions, and have difficulty combining both emotional and cognitive decision making strategies. An fMRI study by van’t Wout and Sanfey (2011) looked at how people with symptoms of schizotypal personalities played the Ultimatum Game, and found that those with these symptoms would both make and accept unfair offers. The goal of this study is to see if schizophrenia participants play the Ultimatum Game differently than those without any psychological disorders. Similar studies have been done with those with schizotypal personalities but the study has not been replicated using actual schizophrenia patients, which must be done in order to fully understand both schizophrenia and how brain activation affects how humans make decisions.

Response to the Ultimatum Game

        The Ultimatum Game is a social decision making game in which there are two players: the first is the proposer, the second is the responder (e.g., Sanfey et al., 2003). The proposer is given a set amount of money, usually $10, and is told to split the money in any way that s/he sees fit. The proposer can offer a range from everything to nothing to the responder. The responder then has one of two choices, accept or decline. S/he can choose to accept the proposition given by the first player. If the responder accepts the offer, then the deal goes through and each player is paid the agreed upon payout and the game is over. If the responder rejects the offer, then neither player gets any money, and the game ends.

        From a purely economic viewpoint, how both players interact is simple. Expected utility (typically denoted as E(U)) is the amount of utility that the player expects from a decision. Utility maximization is the idea that a person will always attempt to get the greatest value out of any particular decision/purchase (Tadelis, 2002). Therefore, the responder should always accept any offer (n) that is greater than 0, because that will maximize utility. If the proposer knows that the strategy of the responder is to maximize utility, then the best response for the proposer is to offer the least amount of money that is not 0, as they knows that the responder will accept. This is not the case in a real life situation. People are affected by their emotions, and if they feel cheated, they will reject offers that they feel is unfair (Gabay, Radua, Kempton & Mehta, 2014). Game Theory has tried to remedy this phenomenon by looking at the perceived percentage chance that a player will behave a certain way. For example, if the proposer in the Ultimatum Game can only make one of two choices, either a fair offer (5, 5) or an unfair offer (9, 1), and he knows with certainty that the responder has a 50% chance to be spiteful (and deny the unfair offer) and 50% chance to be a utility maximizer (who will accept any positive offer regardless of fairness), he can reason that it is better to give the fair offer.  

        Many researchers (e.g., Sanfey et al., 2003) study the behavioral and social effects of the Ultimatum Game by differing the type of opponent from human to computer. The results of the Ultimatum Game against a human are consistent. Fair 5 - 5 offers are almost universally accepted. Seven – three offers are accepted about 90% of the time (Sanfey et al., 2003). 8 – 2 offers are accepted about 50% of the time. 9 – 1 offers are only accepted about 37% of the time. Some researchers, such as Sanfey et al. (2003) feel that by not accepting the unfair offer, the responders are acting to punish the proposer for not practicing the societal expectation of fair play. The responder does not punish nearly as much when playing against a computer opponent. Sanfey et al. (2003) found that when playing against a computer opponent, responders accept 8 - 2 offers about 85% of the time, a full 35% more than when playing against another human. Similarly, 9 – 1 offers are accepted about 60% of the time when against a computer (see Figure 1 for a visual comparison between the two opponent conditions).

        Brain activation can be measured through a device called a functional magnetic resonance imaging (fMRI) machine, which is typically used to look at blood flow to the brain. Hemoglobin are the cells that carry oxygen through the bloodstream to the rest of the body. The hemoglobin has a different magnetic charge when it is carrying oxygen than when it isn’t. The fMRI can then show what parts of the brain are using the most oxygen, giving an idea of which part of the brain is most active for the task at hand.

        When a responder is offered an unfair offer from another human player, there appears to be more blood flow to the bilateral anterior insula (Gabay et al., 2014). The bilateral anterior insula is often associated with negative emotional states, such as pain and disgust, as well as hunger, thirst, and autonomic arousal (Sanfey et al., 2003). In this game, the bilateral anterior insula would respond to the unfair offer with feelings of disgust. Increased activation of this area would most likely result in the responder rejecting the unfair offer (Gabay et al., 2014). The dorsolateral prefrontal cortex (dlPFC) also activates, but for different reasons. This area of the brain is linked to cognitive processes, specifically goal maintenance and planning future behavior (Glimcher & Fehr, 2014). In this game, the dlPFC will act to try to maximize utility, which in this case is to get as much money as possible. These two brain areas are at odds with each other. In fact, when the bilateral anterior insula is more activated than the dlPFC, the responder is incredibly likely to reject an unfair offer. And although activation of the dlPFC is not sufficient enough to predict decision behavior (Sanfey et al., 2003), accepted unfair offers show greater dlPFC activation than anterior insula activation (Gabay, et al., 2014).

        When people play against a computer, as opposed to another human, there is a major difference in brain activation. The bilateral anterior insula does not activate as much as it would if playing against a human opponent, but does activate a little when the computer makes a 9 – 1 offer (Gabay et al., 2014). The explanation is that the responder knows that the computer is not trying to spite the responder, it is only doing what it is programmed to do.

Schizophrenia and Decision Making

        Schizophrenia is a neurodevelopmental disease (Torrey, 2002) that typically affects about 1% of the United States population (Julien et al., 2011). The disease typically manifests during young adulthood, usually late teens to late twenties, although there is some research to suggest that schizophrenia both emerges later in women and have less powerful symptoms than men (Riecher-Roessler & Hafner, 2000). There are both positive and negative symptoms of the disease. The positive symptoms are hallucinations (usually auditory), delusions, and disorganized thoughts and/or speech. Some negative symptoms include emotional blunting, loss of motivation, loss of ability to feel pleasure, and social avoidance (Julien et al., 2011). Additionally, patients with schizophrenia also have trouble paying attention and poor verbal memory (Bowie & Harvey, 2006). Schizotypy is the theory that states that psychosis related disorders (specifically schizophrenia) are on a continuum, similar to the autism spectrum (van’t Wout & Sanfey, 2011). There are several normalized tests, such as the Schizotypal Personality Questionnaire – Brief, which can be used to measure where on the schizotypal spectrum a person lies (van’t Wout & Sanfey).

        A recent study by van’t Wout and Sanfey (2011) looked at how people who scored highly on a schizotypal trait test would reject fewer unfair offers from human opponents in the Ultimatum Game. Additionally, when the Ultimatum Game was given to patients with schizophrenia, an fMRI scan showed that the dlPFC failed to activate in 12 of the 14 participants. Additionally, when those with schizophrenia would play as the proposer, they would occasionally make “ultra-fair” offers, such as 6 - 4, where they make an unfair offer for themselves. Additionally, those with schizophrenia have enlarged ventricles as well as damage to the medial temporal and prefrontal cortex, although researchers are still conflicted on whether this is due to the actual illness or to the anti-psychotic medication typically prescribed in cases of schizophrenia (Torrey, 2002).

Current Study

        The main goal of this study is to look at brain activation differences in schizophrenics and neurotypicals when presented with both a human and a computer opponent in the Ultimatum Game. There is no research on what brain areas will activate in patients with schizophrenia when playing against a computer, and it might give more clues on how schizophrenia affects brain activation when playing the Ultimatum Game.

           To study this, I will use an fMRI machine to look at brain activation in the two participant groups, adults with schizophrenia and adults without schizophrenia, while they play as the responders in the Ultimatum Game against three conditions: a human opponent, a computer opponent, and a control condition. The control condition will be the participant picking between two different amounts of money, just to make sure that all the participants are in fact trying to maximize their utility (which in this case is picking the larger dollar amount of the two offers). The results of their decisions will be compared between participant groups and between opponent groups.

        I hypothesize that the non-schizophrenia group will accept more unfair offers from computer opponents than from human opponents. The participants with schizophrenia will usually accept offers that maximize their utility, regardless of opponent type. In participants without schizophrenia, heightened activity in the anterior insula will be correlated with rejecting an unfair offer. Schizophrenia participants will be more likely to accept unfair offers from human opponents than the non-schizophrenia group. Finally, regardless of the opponent condition, those with schizophrenia will have little to no activation of the dlPFC.

Methods

Participants

        There will be 50 total participants, 25 with no history of psychological disorders and 25 with clinically diagnosed schizophrenia from Strong Behavioral Health in Rochester, New York. All participants will take the Schizotypal Personality Questionanaire – Brief in order to ensure that those in the schizophrenia group do possess these traits and to ensure that those in the non-schizophrenia group do not possess any psychotic traits. The ages of the participants will be from 20 – 30 years (M = 25, SD = 3). Because it has been shown that there are differences in schizophrenia between genders (e.g., Ochoa et al., 2012; Riecher-Roessler & Hafner, 2000), all participants will be male, because it appears that they are more likely to experience more powerful symptoms than females. The schizophrenia participants will be recruited through the Strong Behavioral Health center. The non-schizophrenia participants will be recruited from healthy employees and students at Strong Behavioral Health center.

Materials

        The Schizotypal Personality Questionnaire – Brief (van’t Wout & Sanfey, 2011) will be taken by all participants. A sample tutorial (see Figure 2) on a computer will be given to responders. A Siemens 3.0 T head-dedicated MRI scanner will be used to collect the data, which will be stored on a computer. Nordic Neuro Lab’s Visua-Stim goggles (see Figure 3) will be used to transmit the proposals to the participant. Choices will be recorded using an fMRI compatible selection device. The Brain Voyager software package will be used to analyze the data.

Procedure

        The procedure listed herein is closely based on the study by Sanfey et al. (2003). Participants will first complete a tutorial on a computer that will explain the rules of the Ultimatum Game. Any money that the responder wins will be given to them at the end of the study, and all the proposals total to $10. After they fully understand the game, they will walk into a room where they will meet 10 “proposers”. These proposers will be representative of the local population, with five males and five females (M = 26, SD = 4). Each proposer will state their name and the participants will be told that they would be playing one round of the Ultimatum Game with each of the proposers. Unbeknownst to the responders, all proposals will be produced by a computer with pre-rendered proposals.

        After being set up in the fMRI machine, the responder will play through three different conditions, with the order of the conditions counterbalanced: human, computer, and control. In the human condition, there will be ten rounds with the following proposals in a random order: four $5 offers, two $3 offers, two $2 offers, and two $1 offers. In addition to each offer will be a picture of one of the proposers in order to help ensure that the responder thinks they are playing against a human (see Figure 4).

        The computer trial will be played exactly like the human condition, but instead of a picture of a proposer, there will be a cartoon picture of a computer (see Figure 5). In the control condition, the responder is given the choice of two different money options and must choose to accept the amount that they would prefer. The responders will make their choices with an fMRI compatible selection device. The participants will have 20 seconds to make each choice before the next choice is offered. The entire study should take about 2 hours.

 Data Analyses

        Data from the fMRI will be both processed and analyzed using the Brain Voyager

Software package. A correlation analysis (Pearson correlation with a Bonferroni adjusted significance level) will be used to see if there is a significant difference in brain activation between the dlPFC and the bilateral anterior insula respectively between both participant type and opponent type. Additionally fMRI analysis will be conducted in a way similar to that conducted by Sanfey et al. (2003). A chi-square test of homogeneity will be used to analyze the actual decisions between groups. For example, non-schizophrenia participants will reject unfair offers of $2 more often than schizophrenia participants when playing against human opponents (χ2 = 9.2, 1 df, P = 0.04).  

References

Csukly, G., Polgár, P., Tombor, L., Réthelyi, J., & Kéri, S. (2011). Are patients with schizophrenia rational maximizers? Evidence from an Ultimatum Game study. Psychiatry Research, 187(1-2), 11-17. doi:10.1016/j.psychres.2010.10.005

Gabay, A. S., Radua, J., Kempton, M. J., & Mehta, M. A. (2014). The ultimatum game and the brain: A meta-analysis of neuroimaging studies. Neuroscience and Biobehavioral Reviews, 47, 47549-558. doi:10.1016/j.neubiorev.2014.10.014

Glimcher, P., & Fehr, E. (2014). Neuroeconomics (2nd ed.). San Diego, California: Academic Press.

Julien, R., Advokat, C., & Comaty, J. (2011). A primer of drug action (12th ed.). New York, New York: Worth.

Larquet, M., Coricelli, G., Opolczynski, G., & Thibaut, F. (2010). Impaired decision making in schizophrenia and orbitofrontal cortex lesion patients. Schizophrenia Research, 116(2-3), 266-273. doi:10.1016/j.schres.2009.11.010

Ochoa, Usall, Cobo, Labad, & Kulkarni (2012). Gender differences in schizophrenia and first-episode psychosis: A comprehensive literature review. Schizophrenia Research and Treatment, 2012(916198), 1 – 9. doi: 10.1155/2012/916198

Riecher-Rossler, A., & Hafner, H. (2000). Gender aspects in schizophrenia: Bridging the border between social and biological psychiatry. Acta Psychiatrica Scandinavica Acta Psychiatr Scand, 100(407), 58-62. doi:10.1034/j.1600-0447.2000.00011.x

Sanfey, A. G., Rilling, J. K., Aronson, J. A., Nystrom, L. E., & Cohen, J. D. (2003). The neural basis of economic decision-making in the ultimatum game. Science, 300(5626), 1755-1758. doi:10.1126/science.1082976

Tadelis, S. (2013). Game Theory: An Introduction. Princeton, New Jersey: Princeton University Press.

Torrey, E. F. (2002). Studies of individuals with schizophrenia never treated with antipsychotic medications: A review. Schizophrenia Research, 58(2-3), 101-115. doi:10.1016/S0920-9964(02)00381-X

van 't Wout, M., & Sanfey, A. G. (2011). Interactive decision-making in people with schizotypal traits: A game theory approach. Psychiatry Research, 185(1-2), 92-96. doi:10.1016/j.psychres.2010.05.013

Figure 1: Average acceptance rates of neurotypical responders to an offer dependent on the type of offer that the proposer made from the research of Sanfey et al. (2003). The offers are differentiated by opponent type, to compare differences between the human and computer opponent conditions.

Figure 2: The tutorial will be played on a computer and will detail how the game works. The computer will guide the participant through the rules of the game. In this case, the participant has just played a game where the proposer made a fair offer of $5 - $5. The participant is now playing a game where the proposer offers himself $8 and the responder $2. The participant will pick “yes’ if they want to accept the offer and “no” if they want to reject the offer. The participant will then be asked if they understand they game. If they do not, then they will get two new examples, and the computer will once again explain the game.

Figure 3: An example of the Nordic NeuroLab’s Visua Stim goggles, which will be used to show the Ultimatum Game to the responders while they are in the fMRI machine.

Figure 4: Example of a proposer picture that will be shown to responders in the human opponent

condition.

Figure 5: Computer picture that will be shown to responders in the computer opponent condition.