Can race be erased

1

• Can race be erased?
• The evolutionary psychology of coalitional
  alliances and social categorization
• We are one of the few species capable of
  cooperating in groups of 3 unrelated
  individuals. What cognitive adaptations make
  this possible?
• Office hours
• Cosmides Thursdays, 230-330 Psych East 3808,
  Tuesdays 1230-130 HSSB 1010
• Mrazek Mondays, 1100-100pm, Bldg 429, room 102
  (mrazek_at_psych.ucsb.edu)
• Course website http//mentor.lscf.ucsb.edu/course
  /spring/psyc155
• E-res password collide

2

• In four-day cycles, parties composed of perhaps
  half a dozen males, sometimes with one or two
  females, would move tightly in patrols along the
  edges of their range in every directionI learned
  where they were likely to stop and listen, and
  many times I heard them exchange raucous calls
  with males from the neighoring community. I saw
  how eager they were to embrace and grab each
  other in reassurance when they heard the
  exciting, alarming call of neighbors. I noted
  how, after listening to check that the other
  party was smaller than their own, they would rush
  forward to chase them from half a mile away and
  sometimes they caught a neighbor and attacked
  him. Sometimes they made a mistake and charged
  toward a party that, despite sounding small at
  first, proved large, a situation that led to an
  immediate, confused, and hilarious retreat by the
  invaders back into the heart of their own land.

• Richard Wrangham Wrangham Peterson,
  1996, pp. 13-14

3
What is interesting here?

• Not the violence (most animals use aggression)
• The cooperation
• Fight conflict between 2 individuals
• War (coalitional aggression) conflict between 2
  groups of individuals, each of which must
  coalesce and function as a cooperative unit.
• Rival coalitions US VERSUS THEM

4
Zoologically rare among nonkin

• Social insects (bees, ants) kin selection
  mitigates free rider problems
• Coalitions that include nonkin Humans, chimps,
  maybe dolphins...
• Elephant seal puzzle
• Cognitive complexity?
• Hobbes and coalitions not strict dominance
• What are the cognitive mechanisms that make
  coalitional action possible?

5
Psychology of Us versus Them

• Sherif Robbers cave exp
• Minimal group Ingroup favoritism
• Expectations of indirect group reciprocity
  (Yamagishi)
• Ease of eliciting outgroup derogation,
  discrimination in allocation of resources
• Make allocation zero-sum or
• Add in any element of resource competition
• Sidanius Pratto Social dominance
• Exp econ Collective action/ Public goods games
• cooperating, punishment of free riders

6
Coalitional psychology

• A set of species-typical neurocomputational
  programs that evolved to regulate within-group
  cooperation and between-group conflict in the
  vanished world of our hunter-gatherer ancestors,
  who lived in small bands with many relatives.
• These programs regulate motivations for
  participating cooperatively in coalitions,
  policing their boundaries, and interacting with
  outgroups.
• Still organize much of human social life today.

Tooby Cosmides, 1988 Tooby, Cosmides Price,
2006
7
Social psychologys disturbing result

• An unobtrusive measure of social categorization
  seemed to show that people automatically encode
  the race (sex and age) of individuals they
  encounter, in a mandatory fashion (across social
  contexts and with equal strength).
• Measure relies on patterns in recall errors...

8
What is race encoding?

• Race encoding the tendency to notice and
  remember the race of each new individual we
  encounter
• Form of social categorization
• Encoding race is a pre-requisite for
  discriminating on the basis of race.

9
Race encoding Ingroup favoritism racism?

• If people do encode race automatically and
  regardless of context...and
• If ingroup favoritism outgroup discrimination
  are easy to elicit
• Then
• Will this make racism impossible to eradicate?
• Disturbing implication
• Prompted a 20 year search for contexts that would
  reduce race encoding No success

10
Failed attempts to reduce race encoding

• US and Britain targets black and white
• Targets are discussing race relations vs. a
  race-neutral topic 2
• Subject is black or white. 1,2
• Subject is led to believe that s/he would soon be
  interacting with the targets 2 (a context
  believed to promote the encoding of individuating
  information over category-based information).
• Subject operates under cognitive load or not. 2
• Subject is warned that there would be a recall
  test. 1
• A competing dimension was included (e.g., targets
  differed in both sex and race). 3
• Attention is drawn to targets race (pay
  attention to peoples race), or away from it
  (pay attention to peoples sex). 3

1. Taylor, Fiske, Etcoff, Ruderman, 1978. 2.
Hewstone, Hantzi, Johnston, 1991. 3. Stangor,
Lynch, Duan, Glass, 1992.
11
Solving the puzzle of race encoding

• No contexts found that would reduce race
  encoding
• Suggestion surfaced that the brain contains
  mechanisms designed to encode race
• But this is not plausible from an evolutionary
  perspective.

12
Brain mechanisms designed to encode race are
unlikely to exist

• Our hunter-gatherer ancestors traveled mostly by
  foot. (residential move usually lt 40 miles)
• Rarely if ever encountered individuals of a
  different race
• Selection cannot favor a mechanism designed to
  encode a dimension of the social world that did
  not exist

13
Therefore...

• Race encoding must be a byproduct of machinery
  that evolved for a different adaptive function.
• But what?

14
Byproduct of what?

• Race encoding is not a byproduct of general
  visual perception mechanisms that encode color
  and shape.
• Color is not encoded in an automatic and
  mandatory fashion
• Stangor et al, 1992 Shirt color not encoded at
  all when it has no social significance.
• Attentional factors that affect color and shape
  encoding do not affect race encoding
• Attend to sex, attend to race no effect. Stangor
  et al, 1992
• No prototype effects on encoding

15
Byproduct of what?

• Hypothesis
• Race encoding is a byproduct of mechanisms
  designed for encoding coalitional alliancesone
  component of an evolved, coalitional psychology

16
Elements of a coalitional psychology

• Motivation to cooperate with ingroup members
• Tracking who has contributed what
• Free rider detection ( categorization)
• Punitive sentiments toward free riders
• Facility at solving coordination problems
• Conceptualizing outgroup as social agent with
  intentions (coalitional ToM?)
• Alliance-tracking machinery

Price, Cosmides Tooby, EHB 2002 Tooby,
Cosmides Price, MDE 2006 Delton, Cosmides
Tooby, in prep.
17
Alliance tracking machinery

• (a component of the coalitional system)
• Function To detect who is allied with whom
• Designed to infer alliances from patterns of
  cooperation and conflict
• E.g., standing shoulder to shoulder, support in
  an argument, coordinated threat to third party

18
Alliance tracking machinery

• Problem
• Behaviors that reveal ones coalitional alliances
  dont happen all the time
• Therefore
• Alliance tracking machinery should be designed to
  pick up on any more readily available cue that is
  correlated with patterns of cooperation and
  conflict

19
Cues correlated with patterns of cooperation
competition

• Incidental correlates
• Spatial distribution (e.g., who hangs out with
  whom)
• Accent, manner of speech, idioms
• Gait
• Manner of dress
• Intentional signals
• Salutes, gestures
• Insignia, tattoos, uniforms, clan tartans

20
Result Shared appearance cues often correlated
with alliance behaviors
21
Alliance tracking machineryPredicted design
features

• Activates when exposed to actual patterns of
  cooperation and competition
• Tracks patterns of cooperation competition to
  compute alliances
• When patterns detected, system extracts cues
  correlated with these patterns
• Boosts saliance of any cue predictive of
  coalitional affiliation
• Coalition cues can become activating signals

22
Alliance tracking machinery would produce race
encoding as a byproduct of its design

• We dont live in racially integrated society, so
  there are patterns of cooperation and conflict
  correlated with race
• a shared appearance cue

Asst. Professor, U. Penn
Rob Kurzban, John Tooby, Leda Cosmides, PNAS
2001
23
Is race encoding a byproduct of alliance tracking
machinery?

• Need experiments in which race is not correlated
  with coalition membership
• Two coalitions, each of which cooperates to
  achieve a common goal (2 basketball teams)
• Common goal defeat the other team
• Each coalition has 2 black 2 white members
• So race not correlated with coalition membership
• Had a fight last year now talking about it
• Form impressions of these people

24
Can race be erased?

• Is race encoding a byproduct of alliance tracking
  machinery?
• When race is not correlated with coalition
  membership
• Do people automatically encode coalition
  membership?
• When there is a shared appearance cue, is
  coalition encoded as strongly as race usually is?
• When race does not predict coalition, but other
  shared appearance cues do, does race encoding
  diminish?

25
Who Said What? paradigm

• Targets have a conversation
• Who said which sentence?
• Encoding/ categorization ? inferred from recall
  errors

Taylor, Fiske, Etcoff, and Ruderman (1978)
26
Who Said What? paradigm
You guys started the fight.
27
Who Said What? paradigm
Youve got to be kidding! Your guy threw the
first punch.
28
Who Said What? paradigm
No way! We were playing a clean gameyou all were
fouling.
29
Who Said What? paradigm
30
Who Said What? paradigm

• You guys started the fight.

31
Who Said What? paradigm

• You guys started the fight.

32
Who Said What? paradigm

• You guys started the fight.

33
Shared appearance cue, correlated with coalition
membership

• Shirt color
• not encoded when it has no social significance
• 2 conditions
• Shared appearance cue ABSENT
• Shared appearance cue PRESENT

34
Visual Cue to Coalition Absent
35
Strength of categorization by coalition race
36
Results

• Coalition was encoded even in the absence of a
  shared appearance cue
• When there was a shared appearance cue, coalition
  was encoded as strongly as race usually is.
• When race did not predict coalition, but other
  shared appearance cues did, race encoding
  diminished.

37
Alternative interpretation?

• Was the decrease in race encoding caused by a
  domain-general constraint on attention?
• If that were true, it should be impossible to
  encode two dimensions at very high levels
• Test Coalition and sex

38
Strength of categorization by coalition, race,
sex
39
Race encoding is reversible

• 4 minutes of exposure to an alternative world in
  which race does NOT predict coalitional alliance
  was sufficient to reduce race encoding.
• The brain is not designed to encode race per se.
  It is designed to sift for coalitional alliances,
  and picks up on any perceptual cue that predicts
  who is allied with whom.
• Race encoding is a byproduct of neural programs
  designed for encoding coalitional alliances.

Cosmides, Tooby Kurzban, TiCS 2003 Kurzban,
Tooby Cosmides, PNAS 2001
40
What social alliances count as coalitions?

• Is antagonism between groups necessary to
  decrease race encoding?
• Or
• Are cooperative coalitional alliances sufficient?
  

41
Can race be erased with cooperation alone?

• Dave Pietraszewski ?
• Two charity organizations
• Each cooperates to achieve a common goal
• Another Visual Cue to Coalition provided
• (t-shirt color)
• Will coalition be encoded?
• Will race encoding decrease?

42
Can race be erased with cooperation?

• Members of 2 charity organizations are engaged in
  a conversation
• Content of conversation helps identify coalition
  membership
• Diagnostic paired w/ non-diagnostic sentences
• Non-diagnostic only at recall test
• Two conditions
• Visual cue to coalition ABSENT
• Visual cue to coalition PRESENT (t-shirt color)

43

• Is coalition encoded when the coalitions are not
  in conflict?
• Will race encoding decrease when a visual cue to
  coalition is provided?

44
Encoding in a Cooperative World
?
Shared appearance
Verbal only
Coalition Cue
45
Cooperative ? coalitions
n.s.!
46
Encoding in a Cooperative World
Diagnostic sentences?
Shared appearance
Verbal only
Coalition Cue
47
Mixed sex coalitions
48
Cooperative world Coalition, race, sex
49
Strong evidence for Sidanius argument

• Social categories
• Race and gender are not psychologically
  equivalent!
• Coalitions / race arbitrary set categories
• Gender is not arbitrary
• Automatic encoding, high across social contexts
• Encoding mandatory?

50
What about all female coalitions?

• Prior experiments all targets male
• Same effects with all female targets?
• Charity experiment
• All targets female

51
Cooperative World All female coalitions

• Verbal only

Shared appearance
Verbal only
Cues to coalition membership
52
Cooperative ? coalitions All Female Targets
n.s.!
53
Can Race be Erased in a World of Cooperative
Coalitions?

• Yes
• (perhaps even more effectively)

54
Conclusions Cooperative world exps

• Coalition-tracking psychological mechanisms can
  be engaged without conflict or antagonism.
• Pro-social/cooperative coalitions are sufficient
  to decrease race encoding
• Race encoding is not mandatory
• Further evidence that race encoding is a
  byproduct of alliance tracking machinery

55
Coalition mapping device

• Cue monitoring
• systems

Programs motivating coalitional behavior

• Acts of Alliance
• Patterns of cooperation
• Patterns of competition

Punitive sentiment toward free riders within own
group
Co-variance Detector _____________________________
_____ Coalition set assignment encoding Us
h, k, m Them 1 j, l, n Them 2 p, r, s
Coordinate cooperate with ingroup
members ___________ Outgroup hostility, violence

kus
jthem1 -

• Shared Appearance cues
• Coalitional markers (e.g. attire, grooming)
• Accent
• Phenotypic markers
• etc...

- jthem1
kus
Programs guiding altruistic behavior (e.g.
help) Ingroup favoritism Outgroup indifference
WTRi
Welfare trade-off ratio estimator
56
General Conclusions

• Coalitional psychology evolved to regulate
  within-group cooperation between group conflict
• Ingroup-outgroup literature, social dominance,
  public goods games, punitive sentiments toward
  free riders...
• Alliance tracking machinery designed to compute
  who is allied with whom
• Looks for patterns of cooperation and competition
• Boosts salience of cues correlated with coop/comp
• Race encoding is a byproduct of alliance tracking
  machinery easily reversible by manipulating
  alliance patterns

57
Subject Target Same sex Opposite
sex
58
(No Transcript)
59
What about cooperation among more than 2 people?

• The design of motivational systems regulating
• Coalitional cooperation and conflict

60
More than 2 The problem of cheating

• in repeated 2-person cooperation and exchange, if
  the other person cheats you, you can protect
  yourself by no longer interacting with him/her
• in n-person collective action, this is no longer
  an effective choice to distance oneself from the
  free-rider, one must distance oneself from the
  cooperating group
• solution keep the group, punish the free-riders
• evolved solution irrational punitive
  sentiments against free-riders
• these sentiments are open to opportunistic
  exploitation by coalitions who impose their
  preferred projects for the general good

61
Punitive sentiment as an anti-free rider
psychological device

• Michael E. Price, Leda Cosmides John Tooby
• Evolution and Human Behavior, 23, 203-231 (2002)
• www.psych.ucsb.edu/research/cep/publist.htm for
  article

62
Defensive scenario

• Imagine that a few years from now, the Russian
  people elect a new, warlike dictator who claims
  that Alaska should rightfully belong to Russia.
  Under this dictator, Russia invades and conquers
  Alaska. Theres good evidence that Russia also
  intends to conquer more US territory, in addition
  to Alaska. In response to this invasion, the USA
  declares war on Russia. But because this war was
  unexpected, the USA has allowed its army to get
  relatively small, and it must start drafting U.S.
  citizens in order to have a chance of winning
  this war. How would you feel about this war?
• strongly
  
  strongly
• disagree
  agree
• 1 2 3 4 5 6 7
• Predictor variables
• If the USA won this war, it would be very good
  for me as an individual
• self-interest in group goal
• If I got drafted for this war, I would probably
  agree to serve
• willingness to participate
• Outcome variables
• If a U.S. citizen resisted this draft, Id think
  they should be punished
• punitive sentiment
• If a drafted U.S. citizen agreed to serve in
  this war, Id think they should be rewarded
• pro-reward sentiment

63
Punitive sentiments toward free riders on
collective actions

• What is the adaptive function?
• Labor recruitment?
• Punishing free riders was selected for because it
  encourages more people to particpate in or
  contribute to the collective action
• Eliminating the fitness advantage that free rider
  designs would otherwise have over participant
  designs?
• If advantage remains, free rider designs would
  outcompete designs that cause participation

64
LABOR RECRUITMENT

• Punitive sentiment toward free riders in
  collective action contexts
• If labor recruitment were its adaptive function,
  what predictions would follow?

65
LABOR RECRUITMENT

• Those most likely to benefit from achievement of
  a group goal should differentially act to induce
  others to participate. Self-interest in group
  goal should trigger punitive sentiment.
• Self-interest should independently predict
  punitive sentiment (even after controlling for
  willingness to participate). Encouraging
  self-sacrifice by others provides the largest net
  benefiteven for a free rider.

66
If encouraging participation by others were the
adaptations ONLY function

• After controlling for self-interest, there should
  be no relationship between willingness to
  participate and punitive sentiment.
• Pro-reward sentiment should track punitive
  sentiment.
• Nothing in the problem of labor recruitment
  privileges the carrot over the stick as a means
  of inducing participation.

67
Other labor recruitment predictions

• Punitive sentiment should be sensitive only to
  labor needs, not to free riders per se. Once
  manpower needs are met, the system should be
  indifferent to the prospering of free riders.
• System should be indifferent to whether a
  non-participant is a free rider or not (e.g.,
  exempt people, people who do not benefit from the
  collective action.). Self-interest in group goal
  should trigger punitive sentiment toward any
  non-participant who could help achieve the goal
  by participating.
• Those who contribute anything less than the
  optimal amount should be targets of punishment
  (even if they are contributing at the same level
  as everyone else). (exp econ).

68
ELIMINATING ADVERSE FITNESS DIFFERENTIALS

• Punitive sentiment toward free riders in
  collective action contexts
• Drive free rider out of group
• Turn free rider into contributor
• Not because more labor needed, but because
  eliminates fitness difference
• Releases inhibited productivity
• Since this is inhibited in presence of free
  riders

69
ELIMINATING ADVERSE FITNESS DIFFERENTIALS

• Punitive sentiment toward free riders in
  collective action contexts
• If eliminating free rider fitness advantage were
  its adaptive function, what predictions would
  follow?

70
ELIMINATING ADVERSE FITNESS DIFFERENTIALS

• The individuals own participation should be the
  specific factor that triggers punitive sentiments
  toward free riders. This is because only those
  individuals who contribute are at risk of
  incurring lower fitness relative to free riders.
• The more an individual contributes, the greater
  the adverse fitness differential s/he potentially
  suffers relative to free riders. A sentiment
  designed to prevent outcompetition by free riders
  should key the degree of pun. sent. toward free
  riders to the individuals own willingness to
  participate.
• willingness, pun. sent.

71
Eliminating adverse fitness differentials contd

• Any functional theory predicts that those who
  have an interest in the goal being achieved will
  be more willing to participate. BUT
• Punitive sentiment should track willingness to
  participate, even after controlling for
  self-interest in group goal.

72
If eliminating the free riders fitness advantage
were the adaptations ONLY function

• After controlling for willingness to participate,
  any relationship between perceived benefit and
  pun. sent. should disappear.
• Willingness to participate should predict
  punishment, but not reward.
• When reward induces a free riding underproducer
  to join a collective action, this preserves the
  underproducers relative fitness advantage
  compared to the producer design that is doing the
  rewarding.
• So reward sentiments should not track punitive
  sentiments, especially among those most willing
  to participate.

73
Does willingness to participate predict degree of
punitive sentiment toward free riders?
plt.05, plt.001
74
Does willingness to participate predict degree of
punitive sentiment toward free riders?
plt.05, plt.001
75
What Predicts Punitive Sentiment?What Predicts
Pro-reward Sentiment?

• Partial rs indicate the effect of each predictor
  variable controlling for the other on each
  outcome variable. The cross-over indicates a
  clear dissociation. This suggests that, in a
  collective action context, pro-reward sentiments
  and punitive sentiments are generated by two
  different adaptations.

76
Rational Choice NOT

• RCT People should not punish when costs of doing
  so cannot be recouped. But they do. (exp econ)
• RCT Targets should be people who could own
  level of cooperation to the benefit of the
  rational agent. So punish anyone who contributes
  less than the optimum (even if they contributed
  at the group average). Yet these folks are NOT
  punished. (exp econ)
• RCT Self-interest in group goal should predict
  punitive sentiment. But it does not.
• RCT Willingness to participate should not
  punitive sentiment INDEPENDENT OF expected
  gain (sunk cost fallacy). But it does.
• RCT Reward should track punitive sentiment.
  But it does not.

77
Perhaps rational choice leads you to support
group norms that are in your interest

• RCT Self-interest in group goal willingness
  to participate should punitive
  sentiment only when BOTH are high (to avoid
  advocating your own punishment). But this is not
  the case. (pun. sent. triggered by
  willingness, regardless of self-interest)
• RCT Those who are exempt are free to punish, so
  there should be no willingness-punitive sentiment
  link in those who are exempt (e.g., women). Yet
  there is.
• RCT Those who are willing to participate should
  advocate rewarding participants (they would get
  the reward!). But willingness does NOT predict
  pro-reward sentiment.

78
Representation of groups as social individuals
group status and group identity

• We can and do think of groups as having the
  properties of individuals representationally, a
  group fits into individual argument slot
• the group may want us to do something
• the group may be angry with us
• the group may owe you or you may owe it
• the group may think something
• groups usually have relative statuses
• the group may suffer an insult
• the group may take action, etc.
• but a group has no actual single mind or body and
  no physical existence - only cognitive
  coordination (enforced by self and others)