Open tarngerine opened 7 years ago
Platt formalizes social traps and fences in terms of Skinner reinforcement:
Skinner reinforcement: situation or stimulus S, where a subject emits some behavior B, followed by some reinforcement or result R. R+ is a positive reinforcement (reward), makes the initial behavior B more probable when S occurs. R- is negative reinforcement (punishment), making B less probably when S occurs.
Platt makes a distinction between short-term R (RS) and long-term R (RL)
In these terms, a social trap can be described as: S => B => RS+ => RL-
That is, a stimulus elicits a behavior and there is a short-term reward (encouraging that behavior) but ultimately causes a long-term punishment
A fence can be described as:
S -> B -> RS- -> RL+
That is, a stimulus elicits a behavior and there is a short term punishment (discouraging that behavior), but there would be a long-term reward
"Reversal of reinforcers" can also be applied to traps, where it isn't a comparison of short/long-term, but rather that personal reward/punishment (RP+ or RP-) equates to group punishment/reward (RG- or RG+). e.g. something that rewards the individual is ultimately to the disadvantage of the group.
Three major types of traps/countertraps:
Locked-in aspects of collective behavior: immediate small reinforcements or punishments lead to self-maintaining behavior
Three types of locked-in patterns in collective behavior:
ways out of social traps:
Nested traps: essentially, traps that are intertwined with each other
Social Traps, John Platt, applies cybernetic concepts (e.g. feedback) to wicked problems
---
social traps
definition: "The term refers to situations in society that contain traps formally
like a fish trap, where men or organizations or whole societies
get themselves started in some direction or some set of relationships
that later prove to be unpleasant or lethal and that they see no
easy way to back out of or to avoid."
example: world population problem, where each family may find pleasure and
advantage in more babies; and the problem of competitive consumption
of nonrenewable natural resources.
countertrap
converse of social trap
definition: self-interest prevents individuals from acting in a way that would
be of value to the whole group.
example: kitty genovese murder in nyc in which 30+ neighbors watched a rape &
murder out their windows without calling the police, because then
they'd probably have to testify in court and there's a chance of
the murderer or his friends hunting you down. everyone assumes someoone
else'd call the police.
formalizing skinner
S-B-R sequence
S is stimulus
B is behavior as result of stimulus
R is reinforcement
experimenter introduces S and R; B is supplied by the subject
positive reinforcement (R+) makes B more likely to happen in response to
B in the future
negative reinforcement (R-) makes B less likely to happen in response to
B in the future
often written on two lines, with the experimenter on the bottom and the
subject on the top:
B B
S R S R
applying skinner formalizations to social traps
social traps depend on the difference between the personal / short-term
reinforcements for a given B and teh group consequences or long-term
consequences of that B.
"A social trap occurs, then, when there is an opposition between the highly
motivating short-run reward or punishment, Rs+ or Rs-, and the long-
run consequences, Rl+ or Rl-"
then, a trap can be defined as:
B
S Rs+ ... Rl-
and a countertrap as:
B
S Rs- ... Rl+
three major types of traps tehy found:
one-person traps / self-traps
group traps / missing-hero traps / kitty genovese
commons traps
one-person traps
"The most important subgroup of one-person traps seems to involve the
simple reversal of rein- forcers after a time delay."
examples:
smoking cigarettes, overeating
examples of countertraps:
saving for old age
"A second subgroup of the one-person traps is that in which the problem
is not simple delay, but rather ignorance of the unexpected or
reversed outcome. The fish swimming into the fish trap does not
know that he cannot get out. In the long run, ignorance is as
lethal as evil. This is the case of the man handling a gun who
shoots himself or his friend because he "didn't know it was
loaded.""
"Another subgroup is that of sliding reinf&rcers. These are reinforcers
that change steadily as you go on repeating a behavior, so that
it becomes less and less rewarding and in fact punishing"
an example: drug addiction.
another: "Once, large families with more babies were good for
survival, and they were a delight, but now excessive
babies have turned into an expense and have contributed
to overcrowding for everyone"
missing hero traps
"When group profit, Rg+, is blocked by Rp- for any personal action, we have
the missing-hero trap."
example: mattress falls on the highway, causing cars to slow down and
go around it. no one wants to move the mattress since it'll be
personally dangerous, and by the time you get to it, you can get
past it so you have no incentive to move it any longer.
another: reluctance of anyone to testify against the mafia.
commons traps
Rg- only fallows becuase of hte excessive number of Rp+ practicioners.
similar to everyone's cows grazing a commons, and then the grass running out
for everyone.
"The problemc annot be solved by one or two heroes volunteering not to
graze their cows on the commons, although such a course is frequently
advocated by men of good-will. And the problem is not the result of
any single person doing anything that is unethical or bad, for if the
number of persons involved were kept small, one can imagine that the
collective good would be well served by the sum of all personal Rp+
rewards. It is when the number is excessive that the difficulty
arises."
example: prisoner's dilemma. "This is one of the types of two-person
non-zero-sum games ... Two prisoners have been caught by the police
in some misdemeanor but who are suspected of worse crimes. They are
held incommunicado from each other and each is questioned. The police
offer a pattern of rewards such that if they both "talk" or "defect"
on each other, they get the standard sentence for their crime; if they
"cooperate" with each other, so that neither talks, they get off lightly
for their misdemeanor; but if one talks and the other does not, the
first gets a reward and goes free, while the second gets a doubly
severe sentence. In this situation, the payoff matrix is designed by
the police so that each man benefits by defecting, no matter what his
partner does."
"Generally, in prisoner's dilemma situations, it is found that the opposing
players tend to lock into either steady cooperation or steady
conflict with each other. Which pattern is obtained seems to
depend critically on the outcome—or should we say the "reinforcement"
of the first few plays. Sometimes a pattern of cooperation
is quickly experienced as mutually profitable and is kept.
But if such a pattern is not started early, it seems to be almost
impossible for anyone to continue to cooperate when his opponent
is continually defecting on him and making money at his expense.
It is hard to keep working for Rg+ when the other party's behavior
keeps turning it into Rp- for you."
"As Rapoport (1971) has emphasized, this dilemma and these alternative
outcomes are remarkably parallel to some aspects of international
relations in the non-zero-sum situations of either mutual economic
dependence or mutual nuclear threat. The United States and Canada
have had locked-in cooperation; the United States and Russia
have had 25 years of locked-in hostility and arms escalation."
another example: sell-a-dollar. dinner party game with four rules. first,
bidding starts at a nickel. second, bidding must go up by 5 cents
per bid. third, bidding must not go over $50. fourth, since everyone
will try to get the dollar for so little, the auctioneer gets the two
highest bids, although only the highest bid gets the dollar.
sell-a-dollar has a narrative. the first of two pivotal moments is when bidding
passes 50 cents. then the auctioneer will be getting back more than
the dollar.
the second moment is when bidding passes $1. this is an important
"locked-in" moment. the other person ahs bid a dollar, and you've bid
95 cents. if you raise to $1.05, even if you win, you will lose money.
but if you don't raise, you will lose a lot more. so you raise.
locked-in aspects of collective behavior
three types of locked-in patterns
1. adam smith's "invisible hand" of marketplace
"He used this term to emphasize the absence of any overt or mechanical
causal mechanism in the stabilization of prices or wages around
some median value in a free economic market of competing individual.
A similar invisible hand tends to equalize and centralize the
political parties in systems with majority (rather than proportional
representation) elections."
2. "invisible fist"
"competition of numerous individuals does not produce agreement
on a median value, but instead runs away from the median, with
either escalation or elimination past some point of no return."
3. "invisible chain" (e.g., bane of colin's existence)
"a loop of transactional relationships among two or more people,
forming self-maintaining systems that are sometimes very damaging
and very hard to get out of."
ways to prevent or get out of social traps
1. change the delay to convert long-range consequences into more immediate
ones
example: putting labels on cigarette packages
2. add counterreinforcers, such as social incentives or punishments, to
encourage or discourage behaviors by their immediate Rs+ or Rs-.
example: punitive laws
3. change the nature of the long-run consequence, Rl-.
this happens sometimes e.g. through new inventions or technology.
4. add Rs+ for competing behavior, which will not lead to the bad long-range
consequences.
example: drink a diet cola with saccharine instead of fattening sugar.
5. get outside help in changing the reinforcement patterns of locked-in loops.
example: "The delinquent child gets reinforced for his behavior
by the attention he gets in being scolded, the excitement of
being chased by the police, or the admiration of his friends;
and the teacher, parents, police, friends, and the child are
caught together in an invisible chain of self-maintaining reinforcement
transactions. Tharp and Wetzel tried to find "mediators" — a teacher
or adult friend who could see the child's daily behavior and
give him immediate reinforcers, such as marks in a book for
increased attention or reading, with so many marks entitling
him to extra television viewing or horseback riding on weekends.
As the child's behavior began to change within a few weeks,
teacher, police, and parents changed their attitudes, and his
friends began to admire him for different things (and were reinforced
by the network for their change of values also). At this point,
Tharp and Wetzel, as the outside "therapists," were able to
withdraw because the system had been flipped to a new
self-maitaining mode."
6. set up superordinate authority to present entrapments, to allocate resources,
to mediate conflicts, and to redirect intermediate reinforcement
patterns to more rewarding long-range goals.
example: "a sheriff system with mayors and courts in a western
frontier town all represent something more than just an outside
therapist. They represent the demo- cratic creation of new superordinate
authority able to manage and correct social traps that were leaing
to collective bads."
nested traps
mixed traps and nested traps are hard to solve
example: invisible fist of media promoting violence, locking in to a
multiplication of violent acts and violent individuals in the community.
John Platt, applies cybernetic concepts (e.g. feedback) to wicked problems
http://web.mit.edu/curhan/www/docs/Articles/15341_Readings/Negotiation_and_Conflict_Management/Platt_1973_Social_traps.pdf