"The motivated irrationality of trusting acts: Attributions, cheap talk, penance, contracts, reciprocity, and other causal forces", J. Keith Murnighan, Kellogg School of Management

Rotman School, Organizational Behaviour and HR Management Speaker Series, March 5, 2004, 10 a.m.

These participant's notes were created in real-time during the meeting, based on the speaker's presentation(s) and comments from the audience. These should not be viewed as official transcripts of the meeting, but only as an interpretation by a single individual. Lapses, grammatical errors, and typing mistakes may not have been corrected. Questions about content should be directed to the originator. These notes have been contributed by David Ing (daviding@systemicbusiness.org) at the IBM Advanced Business Institute ( http://www.ibm.com/abi).

Introduction by Mark Weber

• Will be presenting research into trust

[Keith Murnighan]

Family lived in Vancouver for 3 years.

Strong relationships, close ties between Mark, John and Jeff

• Don't do network research, but surrounded by people who are.

Can boil down to a strong tie between Kellogg and Rotman, and weak ties all around.

First will talk to data

• Then will speak to an agenda

Data, how we got started on doing research into trust

• 1995, 3 experimental economists on U. Minnesota, "The Investment Game".
• We call it "the trust game"
• Come into the lab, given instructions, told that others are given instructions (although none are ever seen).
• Given $20.
• • Could keep, or send some to the other person.
  • If you send the money to the other person, it triples.
  • Once they've received the money, they have the choice to give you some, or go home.
• From an economist's perspective, fascinating.
• • If start from the end, to see rationality, the person at the other end say that the other person knows that they'll never see you again, so why would they send you anything?
  • Thus, you should send anything.
• Results (even economics studies) show that not everyone walked.

Suggested follow-up research on this group.

• One group of just senders, other group of just receivers, not connected.
• From an experimental point of view, want to have control over how much is sent (e.g. half, small, large portions).
• How much would you send?

What happens?

• Y-axis, player 1's final outcome.
• Two conditions, people either started with $20 or $10.
• • Sent either $2, $3, $5, $9 or $10
• $5 senders ended up getting $9, i.e. $4 back -- thus worse than not sending.
• Can visually draw a U-shaped curve.
• If you have $20, and send $6 or $9, you do worse than not sending.
• If you have $20, and send $18 or $20, would be better off.
• • However, this doesn't show variance.
  • In a $20 send, 40% of the people got $30.
  • About 30% returned zero.
  • About 20% returned $20.
• The amount you send is the amount you show that you trust the other person.

Today's title:  motivational irrationality ...

• Trying to understand the development of trust.

Interpersonal trust:

• Anyone and their agents.
• Prototypical example:  a teenaged boy meeting an attractive girl, and having to risk revelation or ...

Ad on American television:  6-year old girl in a grass field, and a rhinoceros charging

• The rhino stops short.
• Can't remember what ad that was for -- ineffective ad!
• A perfect definition for trust

Academic literature:

• Mayer, David and Schoorman (AMR, 1995):  the willingness to be vulnerable ...
• Rousseau, Sitknin, Burt & Camerer (AMR, 1995):  trust is a psychological state ...
• Not cooperation, not altruism.

Bayes theorem is a good way to think about old rational models of trust

• Test, see if they'll burn you.
• Thus, in a $20 example, send $2, and see what happens.
• Without a recurring relationship, don't know.

Osgood Graduated Reciprocal Intention Model (GRIM)

• Trust a little, then exponential trust, then flattens off because you can only trust so much.

If there's a negative interaction, then it's costly to rebuild.

• Some believe that once trust is broken, it can't be rebuilt, but this is shown to now be true.

Kramer (Annual Review of Psych 1999) with the rational choice model as the most influential image of trust within organizational science.

• Rational model is parsimonious, ...
• However, curious about those situations that depart for rational model.
• Have been researching "generalized trust":  there's a trust bias, that people are willing to trust strangers to a certain amount
• Russell Hardin says that short term interactions aren't trust, and there's no long term relationship.

First study:  Pillutla, Malholtra & Murnighan JESP, 2003:  Attributions of Trust and the Calculus of Responsibility

• How much return, given the amount sent?
• Experiment 1, in-subject design:  people were given lots of responses, saying that lots of people had sent.
• Also did where people only got one offer.
• Only did the $20 condition.

Theories/hypotheses:

• Game theory:  trust won't be reciprocated.
• Linear model:  tit for tate, Cialdini and obligatoins; desires for equality
• Kramer, Holmes -- in the traditional, rational, model -- suggests small trusting actions will be reciprocated, and large ones are risky.
• Obligation and self-impression management models:  we see ourselves (no one else) and we don't want to see ourselves as a jerk.
• • If someone sends you $5 and had $20 or $10, then are more likely to send back $5 of $10 than $5 of $20.
  • Exponential:  If you send it all, it's a strong signal, and gives the recipient no attribution.
  • • Any less, and the person thinks "why didn't you make the pie bigger", and then can self-justify not returning more.
  • Predicted reciprocity would have the most return.
  • Model of self-impression management in a rational content might return zero.

Found:  sending small amounts led to weak feelings of obligation.

• Returns were fully mediated by felt obligations.

Cultural aspects?

• Haven't run this in Asia, yet.
• Attribution is likely to have cross-cultural different.
• [Study done in China and Japan, not found result different].

In post-experimental questions, people didn't much to be more to send.

• Those who sent more were not perceived as very smart.
• But players who returned more saw these senders as smart.

Now experimenting with notes sent with money.

Early conclusions:

• Feelings of obligation and self-impression
• Send zero or send it all, in between loses.
• • Partial trust is not perceived well.
  • "Partial trust sucks".

Second study:  Org Science 2002, Bottom, Daniels, Gibson and Murnighan, "When Talk is Not Cheap"

• At faculty for advanced study at Stanford
• This group was the trust group, all assistant professors at the time.
• When trust is broken, can it be regrouped?
• Wrote in 1992, submitted to three journals:  revised, resubmit, then rejected (even though was an associated editor at the time!)
• Finally published in Org Science

Emotional concessions are required by both injurer and injured.

• Foregiveness can be cheap talk.
• It doesn't take many in the world to make this bad.

Prisoner's dilemma, where it paid to cooperate (research with undergraduates).

• Half for 5 rounds, half for 15 rounds, two defections.
• After half, decided to make choices at once
• • In short game, this led to immediate defection.
• Acknowledged as intentional or unintentional results.
• Then blasted four more rounds to try to extinguish cooperation.
• Finally, came through with an apology.
• • One condition, simple verbal apology.
  • Another condition as substantive penance, paying back penalty.
  • Open condition:  "what can I do".

Results:

• First five rounds, pretty well everyone cooperated.
• Conventional thinking that once the end is announced it's over, didn't happen.
• Small and large penance didn't make much difference.
• First ran "what will it take", giving no response and no cooperation
• • Then ran "what can I do", these were the most cooperative.
  • Little difference, huge impact.
• Some "what will it take" people, a few asked for a little, but some asked for the world.

New project:  "Getting Off on the Wrong Foot", "Lount, Zhong, Sivanathan, Murnighan

• In four conditions, just say hello; some others say hello later.
• Some defections on first rounds, then cooperation afterwards.
• What hurts the most?
• • Be careful of the middle.
  • Whacking them at the beginning is not so bad, whacking them at the end is not so bad, but whacking them in the middle -- look out!

Malhotra and Murnighan, ASQ 2002, The Effects of Contracts on Interpersonal Trust

• Will a contract help ensure cooperation?
• Sequential prisoner's dilemma.
• People told would play more than once, computerized interaction
• • Contracts were possible:  one immediately binding, or one not immediately binding.
• Every time a binding contract is offered, should take it.
• • But then what happens when a contract is taken away.
• Verbal contract in the simplest way

Hypothesis:

• If there was a contract in place, could either attrribute trust to the person or the contract, and most will attribute to the contract.
• If other party removes contract, attribute the trust to that.

Suggest:

• Binding contracts up front work, but when they're removed, they don't work.
• If you have a binding contract up front is removed exogenously, then trust maintains.

Initial trust between truster and the trusted party is different.

• When trusters' risk are low, they trust more.

Summary:

• Rational choice doesn't always work.

How should attributional models work?

• A Motivated Attrributions Model of Interpersonal Trust:  Mark Weber, Deepak Malhotra and Murnighan
• When the person has $20 and sends $10, rational model would say recipient would understand.
• • However, this isn't true.
  • Sequential process required.
• People as insensitive to other people's payoffs:  disadvantageous versus advantageous equity.
• "We are in the centers of our own universes".
• Assymetry in dependence relationships:  book publishers hate graduate students (who love books); but then love assistant professors (who don't need books).

Suppose two parties, both who would benefit:  who steps first?

• Suggest the one that has more dependence should act first.
• That person might actually risk more than an independent observer thinks warranted, because vision is clouded.

Egocentrism:  may be thinking too much of own risks, not benefits for the other party.

• Appleby, Miller & Rothspan's (1999) gay couples
• • Those who stray from the committed relationship don't practice safe sex
  • Big risks taken earlier -- not rational.
• Fine & Holyfield's (1996) mycological societies ethnography:
• • New members are willing to consume potentially deadly mushrooms, collected, identified and cooked by strangers.
• As dependence goes down, so does trust.

Summary and conclusions:

• Lots of opportunities for empirical and theoretical advances.
• Cheap talk does work; substantive/personal penance works better
• Reciprocating money:
• • Partial trust doesn't pay
  • A lot depends on how we are seen, rather than what we have done.
  • Now researching how we are seen.
• Contracts:
• • Contracts get in the way.
  • The contract formulation process is a way to create trust?
  • Promising to have non-binding contracts.
• People often take risks to establish trust, rather than establishing trust to take risks.