A Self for Others: Joint Self-Other Representation of Value During Morally Relevant Action Remya Nair (rnair@caltech.edu) Division of Humanities & Social Sciences, California Institute of Technology Pasadena, CA 91125 USA Mark Graves (markgraves@fuller.edu) Travis Research Institute, Fuller Theological Seminary Pasadena, CA 91182 USA Kevin S. Reimer (kreimer@uci.edu) Department of Education, University of California Irvine Irvine, CA 92697 USA Warren S. Brown (wsbrown@fuller.edu) Travis Research Institute, Fuller Theological Seminary Pasadena, CA 91182 USA Steven Quartz (steve@hss.caltech.edu) Division of Humanities & Social Sciences, California Institute of Technology Pasadena, CA 91125 USA Gregory R. Peterson (greg.peterson@sdstate.edu) Department of History, Political Science, Philosophy & Religion, South Dakota State University Brookings, SD 57007 USA Dirk Schümann Institute for Systems Neuroscience, University of Hamburg Medical Center - Eppendorf D-20246 Hamburg, Germany Jan Gläscher Institute for Systems Neuroscience, University of Hamburg Medical Center - Eppendorf D-20246 Hamburg, Germany Michael Spezio (mspezio@scrippscollege.edu) Department of Psychology, Scripps College Claremont, CA 91711 USA Institute for Systems Neuroscience, University of Hamburg Medical Center - Eppendorf D-20246 Hamburg, Germany Division of Humanities & Social Sciences, California Institute of Technology Pasadena, CA 91125 USA Abstract in groups of people who 1) demonstrate real-world stable and reasoned action for others in long-term commitments of The cognitive science of moral action seeks accounts of moral compassionate care; 2) demonstrate stable and reasoned cognition – and their conceptual and valuational structures – action in the laboratory over 2-3 years and across context; and that explain stable or unstable, reasoned or unreasoned, moral 3) a large group of young adults. We compared 4 different commitments in the real world. To be successful, cognitive models, intended to correspond with being insensitive to science requires experimental approaches that are relevant to context (Model 1), with simple ethical utilitarianism (Model the lives and choices of people who demonstrate stable moral 2), with an ethics of nondual self (Model 3), and with an commitment in real life. Further, cognitive science should be ethics of relationally nondual self (Model 4). In all 3 studies, able to develop models analogous to the theories from other greater action for others associated with having a joint scholarly inquiries into moral cognition, such as moral representation of values for self and others while still philosophy and theology. We applied cognitive valuational differentiating between the two (Model 4). Our findings show modeling and Bayesian model comparison to analyze choices that action for others is facilitated by having a “self for 330 others”: a representation of value for self that is tied to value laboratory group understands calculus or the language under for others without losing the distinction between the two. investigation, respectively. The results from such work would assuredly include replicable patterns across groups, Keywords: moral action, moral cognition, virtue, character, decision science, valuational modeling, Bayesian model but these patterns would likely be unhelpful to the comparison understanding of the cognitive processes calculus and language. Introduction We took two approaches to avoid moral averaging. In Study 1, we worked with a group of participants with long- Two primary questions motivate this research. The first is term commitments to stable and compassionate care of whether the application of cognitive modeling adults with mild to profound neurodevelopmental disorders, methodologies from decision science, using discrete choice primarily via close, one-to-one caring dyads. They are all theory (W. H. Greene, 2009; Mazzanti, 2003), Bayesian members of the L’Arche organization (L’Arche USA; parameter estimation (Kruschke, 2010), and Bayesian model http://www.larcheusa.org/), which was recognized by Pope comparison (Gelman, Hwang, & Vehtari, in press; Vehtari John Paul II as “a sign of hope in a divided world.” We then & Gelman, 2014) could reveal how people value self and applied cognitive valuational modeling to the choices that other during contexts that allow, but that do not require, they made in a novel “rescue decision” task that allows, but compassion and costly care for others. Do people who act does not require, costly care for others under ambiguous more often for others represent the value of self and of other threat to self. In Study 2, we first used decisions about the differently than those who choose not to care when the common good to classify a people as Giving and as matched opportunity arises? Do people who care have a cognitive Controls. We tested the stability and the generality of both representation of joint valuation that is absent or simply the Giving and Control groups by asking them to return modulated in people who act less often on behalf of others? after 2-3 years to complete the same “rescue decision” task We compared four different cognitive valuational models as the L’Arche members had done. We also assessed intended to correspond to the following broad ethical whether those who stably acted on behalf of others theories: 1) context insensitivity, which might include a displayed a valuational representation similar to the L’Arche deontological ethics (Herman, 2007; Kant, 1996 (1798), sample. In Study 3, we applied the valuational modeling 2005); 2) simple utilitarian ethics, with its focus on outcomes from Studies 1 and 2 to an analysis of data from a additivity in aggregates of value (J. D. Greene, Nystrom, large group of young adult participants in a northern Engell, Darley, & Cohen, 2004; Mill, 1871); 3) an ethics of European city, to determine whether valuational nondual self, in which the values of self and other are representations from the two groups of stable givers in merged such that one is indistinguishable from the other Studies 1 and 2 would also associate with giving in a more (Gethin, 2011; Heim, 2011; Lopez, 2008); and 4) an ethics generic group of participants from a different cultural of relationally nondual self, in which the values of self and context. All studies involving the Rescuer Paradigm (RP) other are held together while maintaining the distinction used real money, and participants always began with twice between them (Aquinas, 1964; Aristotle, 1992 (1925); as much money as the Victim. In Studies 1 and 2, the Bonhoeffer, 1998, 2005(1949); Frick, 2008). Victim is a real person who is not present and who is The second major question is whether laboratory tasks unknown to the participant and who will never have an designed to study ethical action are useful for groups of opportunity to reciprocate any help that the participant people whose long-term decisions in the real world show provides. In Study 1, participants began with a total of clear evidence of costly action on behalf of others. US$60, in Study 2, participants began with a total of Economists, educators, policy makers, religious leaders, and US$90, and in Study 3, participants began with a total of 15 grant-making foundations have all raised serious doubts euro. about whether tasks designed in the laboratory, sometimes dismissed pejoratively as “only games”, are able to meet this challenge. Moreover, unless such an extension of Study 1 laboratory tasks is possible, even scholars who are open to In the first study, 48 members of L’Arche USA (Age: M ± cognitive scientific approaches will object that the research SD = 40.9 ± 15.9, range = 21-84 years, 34 women) findings from laboratory participants are unhelpful due to completed 30 rounds of the Rescuer Paradigm (RP; see “moral averaging.” (Peterson, Van Slyke, Spezio, Reimer, Figure 1), in which on each trial a participant observes a & Brown, 2010) perpetrator steal money from a victim ((Spezio, Brown, “Moral averaging” refers to the practice of making Peterson, Reimer, & Van Slyke, 2008); Figure 1). Briefly, theoretical and mechanistic inferences about moral on each round, the Participant (Observer) witnesses a cognition from measures during ethically salient choices Perpetrator stealing money from a Victim. The Participant made by typical laboratory participants, whose actual has the option of helping or not helping the Victim. To help histories of unstable, stable, reasoned, or unreasoned the Victim, the participant gives of her/his own money to ethically salient choices are unknown. This practice is akin make up for the amount stolen. Each time the participant to developing a cognitive science of calculus or a cognitive helps the Victim, the probability that the Perpetrator will science of language without first assessing whether the detect the participant increases. If the Perpetrator detects the 331 Participant, the Perpetrator steals all of the Participant’s the Victim: 1) Model 1, a model that included no contextual money. Thus, the RP tests participants’ willingness to take variables such as the amount stolen or how much the action for others within a context of ambiguous participant or the Victim had lost (i.e., Contextual threat/danger. Insensitivity); 2) Model 2, an additive model that separated loss to self from loss to the Victim as two predictor variables (i.e., simple Utilitarian); 3) Model 3, a model that tested for a multiplicative combination of loss to self and loss to Victim that lost all distinction between them (i.e., Nondual Self); and 4) Model 4, a model that offset loss to self and loss to Victim in a unified ratio (i.e., Relationally Nondual Self). RStan (RStan_Development_Team, 2014; Stan_Development_Team, 2014) generated Bayesian parameter estimates and model comparison used the WAIC statistic (Gelman et al., in press; Vehtari & Gelman, 2014; Watanabe, 2010). Almost all of our participants (78%) Figure 1: The Rescuer Paradigm. Participants observe favored some form of joint self-other value representation, money being stolen from an anonymous Victim and are with 63% showing a ratio (Model 4) and 15% showing a given the choice of whether or not to help. multiplicative (Model 3) representation of joint value. Those who represented joint value as a ratio (Model 4) also gave Study 1: Results more to the Victim (R2 = 0.77). L’Arche USA members gave nearly 60% of the time, with half giving above this proportion of trials (Figure 2A). Study 2 Several gave on every trial, and these are shown In the second study, 203 participants from the greater Los overlapping in the circle at the very top of the graph. Angeles area completed one session of the Public Goods Reports by the participants are consistent with the view that Paradigm (PGP) in groups of 10-12 people (15 rounds, $10 this level of giving was intentional and rational according to initial endowment per round which could be doubled to a the participants’ own value judgments. For example, one of $20 payout for the entire group if at least 25% of the group the participants reported giving the maximal amount on contributed). We defined a Giving group by selecting all every other trial, so as to balance a lower probability of participants who gave on at least 13 of 15 rounds, including detection with a high degree of caring action for the Victim. the first and last round (N=17). Of the people who never The loss ratios (loss to self divided by loss to other) prior to gave or gave on at most 1 round, a Control group (N=17) any detection events were greater than 1 for nearly all matched the Giving group on self-reported age, gender, participants (Figure 2B), indicating more choices to give of income, big 5 personality, empathy, and prosocial one’s own money than to allow the Victim to lose money. personality. After 2-3 years, participants from both groups returned to the laboratory to complete a 15-round Rescuer 1 10 Paradigm. 9 0.9 0.8 8 Study 2: Results 0.7 7 After a delay of 2-3 years following group classification according to behavior on the PGP, participants in the Giving 0.6 6 group gave a higher proportion of their money to the RP 0.5 5 Victim (M ± SD = 0.45 ± 0.1; Figure 3A) compared to the 0.4 4 Control group (0.08 ± 0.06; Figure 3B), demonstrating 0.3 3 stable morally relevant decision making across behavioral 0.2 2 1 contexts and extended periods of time. Cognitive 0.1 0 0 valuational modeling followed by predictive model fitting using WAIC showed that most of the Giving group, but not Figure 2. A (left). Proportion of “give” choices across the the Control group, jointly represented losses to self and 30 trials, plotted for all participants. Horizontal bar is the other as a unified ratio (Model 4). median. B (right). Loss Ratio ([LossSelf]/LossVictim]), plotted for all participants (3 points fall below a Loss Ratio of 1, shown by the dotted line) Horizontal bar is the median. To determine how the L’Arche members represented their own losses with respect to the losses of the Victim, we compared four different logistic regression models in terms of fit to the trial-by-trial choices to keep or to give (0 v. 1) to 332 Conclusions Our findings show that action for others is facilitated by a “self for others,” that is, a representation of value for self that is tied to the value for others, but that preserves a distinction between them. This finding opens up Figure 3: Trial-by-trial giving on the RP following a 2-3 possibilities for interdisciplinary inquiry with deontological year delay from the initial group classification on the PGP. theories, with utilitarian theories, with theories emphasizing A (left). Giving group. B (right). Control group. nondual self, and with theories holding to a relationally nondual self. We also show that laboratory methods and Study 3 cognitive valuational modeling are relevant to In the third study, 503 young adult participants recruited understanding the moral cognition serving stable moral from Hamburg, Germany, and the surrounding area commitment in real life. Once this relevance is established completed a 15-round Rescuer Paradigm as part of a large and once the exemplary patterns of valuational study of learning and decision making. The battery of representation are identified, those same methods can be testing materials included the DOSE assessment of risk and applied to populations more generally. loss aversion (Wang, Filiba, & Camerer, 2010), the Portrait Values Scale (Schwartz, 2006; Schwartz & Boehnke, 2004), and the Temperament and Character Inventory (Cloninger, Svrakic, & Przybeck, 1993). Acknowledgments We are grateful to Kathryn Aughtry, Andrea Beckam, and Study 3: Results Catherine Holcomb for assistance in data collection and to Only 413 of 503 participants chose to give to the Victim on James Van Slyke, Kristen Monroe, and Linda Zagzebski for at least 2 of 15 rounds. We found that self-report ratings helpful discussions. We gratefully acknowledge funding associated only weakly with actual behavioral outcomes from the Science and Transcendence Advanced Research such as the proportion of giving or the loss ratio. We Series of the Center for Theology and the Natural Sciences determined how participants represented their own losses in Berkeley, CA, from the Center of Theological Inquiry in and the losses to the Victim by again using the WAIC Princeton, NJ, from the John Templeton Foundation (Grant criterion compare each of the four models of their trial-by- 21338), and from the German Ministry of Research and trial choices. Most of the participants’ choices were not fit Education to J.G. (01GQ1006). best using a model that represented loss to self and Victim as a joint value. However, participants showing References representation of joint value as a ratio of loss (Model 4) also showed a greater propensity to give to the Victim (F(3,409) Aquinas, T. (1964). Commentary on Aristotle's = 8.68, p < 0.0001), compared to the context-free model Nicomachean Ethics (C. I. Litzinger, Trans.). Notre (Model 1; z = 3.02, p < 0.05); the additive model (Model 2; Dame, IN: Dumb Ox Books. z = 2.00, p < 0.05), and the multiplicative model (Model 3; z Aristotle. (1992 (1925)). Nicomachean Ethics (W. D. Ross, = 4.34, p < 0.05). Trans.). New York: Oxford University Press. Bonhoeffer, D. (1998). Sanctorum Communio: Theological Discussion Study of the Sociology of the Church (R. Krauss & N. We applied cognitive valuational modeling of choices on a Lukens, Trans. Vol. 1). Minneapolis, MN: Fortress. “rescue decision” task, completed by two groups of Bonhoeffer, D. (2005(1949)). Ethics (I. Todt, H. E. Todt, E. participants demonstrating stable, caring action for others. Feil & C. Green, Trans. Vol. 6). Minneapolis, MN: The first group’s stability of care came in the form of their Fortress. long-term commitment to supportive relationships with Cloninger, C. R., Svrakic, D. M., & Przybeck, T. (1993). A adults who have neurodevelopmental disorders. Evidence psychobiological model of temperament and character. for stable care in the second group came from measures Archives of General Psychiatry, 50(12), 975-990. over 2-3 years in different morally salient laboratory tasks. Frick, P. (2008). The Imitatio Christi of Thomas à Kempis We found that when participants in both groups represented and Dietrich Bonhoeffer. In P. Frick (Ed.), Bonhoeffer's value jointly as a ratio of self and other (Model 4), action for intellectual formation : theology and philosophy in his others increased. Further, in a larger sample of young adults, thought (pp. 31-52). Tübingen: Mohr Siebeck. we also saw that this ratio representation of joint value Gelman, A., Hwang, J., & Vehtari, A. (in press). associated with a greater propensity to act on behalf of Understanding predictive information criteria for others, compared to models that represented value Bayesian models. Statistics and Computing. separately or multiplicatively. 333 Gethin, R. (2011). On some definitions of mindfulness. http://www.stat.columbia.edu/~gelman/research/unpublis Contemporary Buddhism, 12(1), 263-279. doi: hed/waic_stan.pdf 10.1080/14639947.2011.564843 Wang, S. W., Filiba, M., & Camerer, C. F. (2010). Greene, J. D., Nystrom, L. E., Engell, A. D., Darley, J. M., Dynamically optimized sequential experimentation & Cohen, J. D. (2004). The neural bases of cognitive (DOSE) for estimating economic preference parameters. conflict and control in moral judgment. Neuron, 44(2), http://people.hss.caltech.edu/~sweiwang/papers/DOSE.pd 389-400. f Greene, W. H. (2009). Discrete Choice Modeling. In T. C. Watanabe, S. (2010). Asymptotic Equivalence of Bayes Mills & K. Patterson (Eds.), Palgrave Handbook of Cross Validation and Widely Applicable Information Econometrics (Vol. 2, pp. 473-556). New York, NY: Criterion in Singular Learning Theory. Journal of Palgrave MacMillan. Machine Learning Research, 11, 3571-3594. Heim, M. (2011). BUDDHIST ETHICS: A Review Essay. Journal of Religious Ethics, 39(3), 571-584. Herman, B. (2007). Moral Literacy. Cambridge, MA: Harvard University. Kant, I. (1996 (1798)). The Metaphysics of Morals (M. J. Gregor, Trans.). In M. J. Gregor (Ed.), Practical Philosophy (pp. 355-603). New York, NY: Cambridge University Press. Kant, I. (2005). Groundwork for the Metaphysics of Morals (T. K. Abbot, Trans.). Orchard Park, NY: Broadview Editions. Kruschke, J. K. (2010). Doing Bayesian Data Analysis: A Tutorial with R and BUGS. Burlington, MA: Academic Press. Lopez, D. (2008). Buddhism and Science: A Guide for the Perplexed. Chicago, IL: University of Chicago. Mazzanti, M. (2003). Discrete choice models and valuation experiments. Journal of economic studies, 30(6), 584-604. Mill, J. S. (1871). Utilitarianism. London, UK: Longmans, Green, Reader & Dyer. Peterson, G., Van Slyke, J., Spezio, M. L., Reimer, K., & Brown, W. S. (2010). The Rationality of Ultimate Concern: Moral Exemplars, Theological Ethics, and the Science of Moral Cognition. Theology and Science, 8, 139-161. RStan_Development_Team. (2014). RStan: the R interface to Stan, Version 2.5. Retrieved from http://mc- stan.org/rstan.html. Schwartz, S. (2006). Value orientations: measurement, antecedents, and consequences across nations. In R. Jowell, C. Roberts, R. Fitzgerald & G. Eva (Eds.), Measuring Attitudes Cross-Nationally: Lessons from the European Social Survey. London: Sage. Schwartz, S., & Boehnke, K. (2004). Evaluating the structure of human values with confirmatory factor analysis. Journal of Research in Personality, 38(3), 230- 255. Spezio, M. L., Brown, W. S., Peterson, G., Reimer, K., & Van Slyke, J. (2008). Virtuous Decisions: Exemplarity in and out of the laboratory. Paper presented at the Society for Neuroeconomics, Park City, Utah. Stan_Development_Team. (2014). Stan: A C++ Library for Probability and Sampling, Version 2.5.0. Retrieved from http://mc-stan.org. Vehtari, A., & Gelman, A. (2014). WAIC and Cross- Validation in Stan. 334