Sacred values & consequentialism Will Bennis Northwestern University Department of Psychology MURI Meeting, 1/27/2007 Collaborators Doug Medin Rumen Iliev Sonya Sachdeva Dan Bartels Outline 1. Sacred / protected values associated with moral rules and lack of concern with consequences. Contrary evidence suggests sacred values—and associated moral rules—may be associated with greater concern with consequences. Some reasons why this makes sense: 2. 3. – – 4. Unrealistic closed-world assumptions Study participants’ use of relevant external knowledge. A challenge: modeling the relationship between moral rules and consequences Sacred & Protected Values • Sacred / protected values (SVs) are values that resist tradeoff no matter what the (secular) benefits (Fiske & Tetlock, 1997; Baron & Spranca, 1997). • While some SVs are widely shared across cultures (e.g., innocent human life), others may be individual or culture specific (e.g., which lives count as innocent). One explanation for refusal to tradeoff SVs • From Baron & colleagues: – SVs are not about consequences, – They are about moral rules: duties to act—or, more often, not to act—in a certain way. – E.g., “Do no harm.” • If true, this is a potential problem for theories of rational choice: – People may often fail to maximize benefits relative to costs, – But they should at least want to. Example from field research in northeast Wisconsin • Participants: Members of various cultural groups in two neighboring counties, including – Native Americans living on the Menominee reservation, – Evangelical Christians living in neighboring Shawano county, – Avid hunters and fishers (in both locations). Scenarios 1. Requiring Native Americans to give up tribal regulation of fishing and hunting practices. 2. Allowing public schools to teach secular (non-religious) evolutionary theory in science class, but not a Christian perspective on creation. 3. Allowing the mother and family to decide whether or not to have an abortion in cases where she would almost certainly lose her life delivering the child. 4. Allowing farmers to use fertilizers that pose a very small risk of groundwater contamination if they dramatically increase the yield of a large field. Participants asked to indicate agreement with 2 statements 1. “This is the kind of decision where it's best to rely on moral rules of right and wrong” 2. “This is the kind of decision where it's best to [weigh the costs & benefits/pros & cons]” 7 pt. Likert scale, 7 = completely agree Average across scenarios (All four scenarios had the same cross-over pattern). “This is the type of decision where it’s… 7 6 5 SV="No" 4 SV="Yes" 3 2 1 "This is the type of decision where it's best to rely on moral rules" p < .001 "This is the type of decision where it's best to weigh the costs & benefits" But… • Conflicting findings (Bartels & Medin, in press; Connolly & Reb, 2003): – Depending on how stimuli are constructed, – People with sacred values appear more instead of less concerned with consequences. One possible explanation: • Under certain conditions, participants may believe that – – • These conditions may be associated with sacred values (i.e., cases where actions are judged unacceptable no matter how great the benefits). If true… • – – – • Relying on moral rules is reliable way to achieve good consequences, Whereas weighing anticipated costs & benefits is not. Disagreement with the statement “…best to weigh the costs & benefits” & agreement with the statement “…best to rely on moral rules of right and wrong” ≠ “I do not care about consequences.” I’ll discuss why this might be make sense later in talk. Two sources of suggestive evidence… …that people with SVs believe relying on moral rules helps them improve consequences: 1. Relationship between the endorsement of two statements: a. “This is the kind of decision where it’s best to rely on moral rules of right and wrong” b. “This is the kind of decision where relying on moral rules of right and wrong will lead to better long-term consequences.” 7 6 5 SV="No" 4 SV="Yes" 3 2 1 "This is the kind of decision where it's best to "This is the kind of decision where relying on rely on moral rules of right and wrong." moral rules of right and wrong will lead to better long-term consequences." r = 0.74 2) Endorsement of 2 statements as to when/why to follow moral rules: a. “The moral rules of right and wrong that I endorse should be followed even if I knew in some particular case that they would lead to worse long-term consequences.” b. “I endorse the moral rules of right and wrong that I do specifically because I believe they lead to better long-term consequences. If I knew they did not lead to better long-term consequences, I would no longer endorse those moral rules.” 7 4 1 Moral rules even if worse Moral rules because better long-term consequences. long-term consequences. • Mean response • 7 point Likert scale (7 = completely agree). • p=.001 by two-tailed t-test. Why might Ps think relying on moral rules works better than weighing costs & benefits? • Decision researchers (including ourselves) tend to make a number of “closed-world” assumptions: – That Ps will and should limit the information they use to that provided in the scenario itself (i.e., they will not use experience or knowledge they bring to scenario). – That Ps will accept the information provided as certain, even if it couldn’t realistically be. – That Ps will not add constraints to unconstrained variables, even if those variables would always be constrained in a real-world decision. Participants often did not accept these closed-world assumptions -1 • In response to the sacred value measure, “[X] is unacceptable no matter how great the benefits”: – many Ps took “benefits” to be constrained by realistic possibilities of the scenario. Participants often did not accept these closed-world assumptions -2 • When considering “weighing costs & benefits,” Ps often: – Included pre-existing assessments of costs and benefits, – Limited “costs & benefits” to secular costs & benefits in accordance with common connotations of the term, – And recognized that the anticipated costs & benefits resulting from the scenarios were more or less certain depending on characteristics of the scenario. Participants often did not accept these closed-world assumptions -3 • When considering the value of “relying on moral rules”: – Ps often included the moral rules’ dependence on past experience – Or on external sources of knowledge outside the scenario, such as their religion or tradition. • Given our closed-world assumptions: Weighing anticipated costs & benefits may always lead to better consequences than following moral rules. • But given our participants’ use of pre-existing experience & knowledge: A preference for relying on moral rules over weighing anticipated costs & benefits may be adaptive if limited to selective situations. Not meant to suggest Ps will always be conscious of connection between moral rules & consequences • • Learned associations between rules and consequences are not necessarily available at conscious/explicit level. For example, the rules may be learned: • • • Through implicit/emotional experiential learning, By other members of community / ancestors & socialized as moral rules, Through evolutionary mechanisms (e.g., tit for tat). A challenge for modelers: • Can we model the relationship between 1. Study participants’ preference for “relying on moral rules” or “weighing costs and benefits,“ 2. Different learning / decision-making processes • • • • Weighing anticipated costs and benefits Individual trial & error rule learning Group level trial & error rule learning Evolutionary rule learning 3. The structure of the task environment (given these scenarios as they are faced in practice by the research population). E.g.: • Estimates of the degree to which costs and benefits can be anticipated, • The time-span across which consequences might occur, • Stability versus rates of change in the task domain. • In order to compare anticipated performance of “relying on moral rules” versus “weighing anticipated costs & benefits”? End … It seemed like a good idea at the time.