liking and learning. neuroscience and paternalism - c. f. camerer

8/10/2019 Liking and Learning. Neuroscience and Paternalism - C. F. Camerer

1/25

Wanting, Liking, and Learning: Neuroscience and PaternalismAuthor(s): Colin F. CamererSource: The University of Chicago Law Review, Vol. 73, No. 1, Symposium: Homo Economicus,Homo Myopicus, and the Law and Economics of Consumer Choice (Winter, 2006), pp. 87-110

Published by: The University of Chicago Law ReviewStable URL: http://www.jstor.org/stable/4495545.

Accessed: 02/10/2014 01:48

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at.http://www.jstor.org/page/info/about/policies/terms.jsp

.JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of

content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms

of scholarship. For more information about JSTOR, please contact [email protected].

.

The University of Chicago Law Reviewis collaborating with JSTOR to digitize, preserve and extend access to

The University of Chicago Law Review.

http://www.jstor.org

This content downloaded from 190.232.56.220 on Thu, 2 Oct 2014 01:48:47 AMAll use subject to JSTOR Terms and Conditions
http://www.jstor.org/action/showPublisher?publisherCode=uclrhttp://www.jstor.org/stable/4495545?origin=JSTOR-pdfhttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/stable/4495545?origin=JSTOR-pdfhttp://www.jstor.org/action/showPublisher?publisherCode=uclr


2/25

Wanting,

Liking,

and

Learning:

Neuroscience and

Paternalism

Colin

F

Camerert

A

large

number of

young

children die

every year

from

ingesting

poisons by

accident.

Revealed-preference

heory,

he

foundation

of

mi-

croeconomics,

has

only

two

concepts

to

explain

these

accidents: eliefs

and

preferences (that

is,

utilities revealed

by choices).

In

the

belief-

preference

anguage,

all we can

say

is that a

child who died of

acciden-

tal

poisoning

either believed the

poison

was

safe,

and

made a mistake,

or

that the child

preferred

death to

life,

and

committedsuicide.While

economic

theory

may

be comfortable

calling

these accidents

only

mis-

takes or

expressions

of

preference,

am

not.

My goal

in

this

Essay

is to

acquaint

the

legal

audience

with ideas

emerging

in

neuroscience that

could

potentially

be

a

richer

anguage

for

talking

about cases like ac-

cidental

child

poisoning

and,

more

broadly,

about

welfare and

pater-

nalism

in

some limited cases. Further

development

of the framework

could lead to a broaderview with wider

applicability.

The idea is that three separableneuralsystemsare relevant for

choice and welfare: a hedonic

"liking" ystem

(welfare),

a

"wanting"

system

that

guides

choice,

and a

learning system

that,

ideally,

links

information tored

in

the other

systems

so

that

people

choose to learn

what

they truly

like. Other

multiple-process

neuroeconomic

accounts

use different

components

but have similar

mplications.'

t

Rea

A. and Lela G. Axline Professor of Business

Economics,

California nstituteof

Technology.

hanks o

conference

participants,

o

Meghana

Bhatt,

Ming

Hsu,

and Ian

Krajbich

for researchassistance, nd to Alice Linfor a presentation n theBerridgework.

1

See Drew

Fudenberg

nd David

K.

Levine,

A

Dual

Self

Model

of Impulse

Control

Aug

17,

2005),

online at

http://post.economics.harvard.edu/faculty/fudenberg/papers/dualself.pdf

vis-

ited Jan

17,

2006)

(arguing

hat a

simple

"dual-self'

model

explains mpirical rregularities

n

self-

control

problems

nd can

give

a value for commitment

n

decision

problems);

sabelleBrocasand

Juan

D.

Carillo,

The Brain as a Hierarchical

Organization Aug

2005),

online at

http://www-

rcf.usc.edu/-juandc/PDFpapers/wp-brain.pdf

visited

Jan

17,

2006) (constructing

model of con-

sumption

and

time allocation hat is solved

using

ools frommechanism

esign

and economicsof

information);

.

Douglas

Bernheimand Antonio

Rangel,

Addiction

and

Cue-Triggered

ecision

Processes,

4

Am

Econ Rev

1558,1561

2004)

proposing

modelof addiction hat s

premised

n a

pathological

ivergence

etween

choiceand

preference

onsisting

f

mistakes,

nvironmental

ues,

and

attempts

by

users

to minimize heir

mistakes);

George

Loewenstein

and Ted

O'Donoghue,

AnimalSpirits:Affectivend Deliberative rocesses n EconomicBehavior July27,2004),onlineat

http://gsbwww.uchicago.edu/research/workshops/behavioral/lowenstein.pdf

visited

Jan

17,

2006)

(discussing

model of humanbehaviorbased

on the interactionof a deliberative

ystem

and an

affective

system);

George

Loewenstein,

Out

of

Control:Visceral

nfluences

on

Behavior,

65

Org

87

http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp


3/25

88 The

University f Chicago

Law Review

[73:87

The

Essay

is

organized

n

several

parts.

PartI sets the

historical

stage

and then describesa canonical

experiment

(with mice)

on

want-

ing

and

liking.

Part II describes anecdotal evidence of

wanting-liking

gaps

from

psychological

disorders and choices over time

(addiction

and credit card

use).

Paternalism

enters

in

Part

III,

not with

much

vigor,

but as

a

potentially

efficient

response

to

gaps

in

wanting

and

liking.

Part

IV

suggests

some directions

for

positive analysis.

PartV

concludes.

I. REVEALED

PREFERENCES

AND

WANTING-LIKING

A.

History

of

Thought:

The Revealed-Preference

Approach

Is a

Convention

In

the

early

1900s,

VilfredoParetowon a

debate

among

economists

about the

epistemological

basis of

utility,'

a

victory

later cemented

in

the 1930s

by

the ordinalist evolution.

Pareto and

his

contemporaries,

uch as

Edgeworth

and

Jevons,

all

believed

in

the "concrete

deductive method" of John

Stuart

Mill,

which starts with

simple empirical regularities

or "laws"

and makes

deductions from those

regularities.3

Furthermore,

ll of these scholars

believed that

understanding

he hedonic basis of

utility

was

necessary,

in a form that is now calledpsychology(butwas not calledpsychology

then).

As

Jevons wrote:"But it is

surely

obvious

that economics does

rest

upon

laws of human

enjoyment;

and

that,

if

those laws are

devel-

oped by

no other

science,

they

must be

developed by

economists....

[T]he theory

of

economics must

begin

with

a

correct

theory

of

con-

sumption."

Pareto's view was the

opposite:

he

thought rooting

economic

preferences

in

psychological

detail

was an

unnecessary

burden.

In

an

1897

letter Pareto wrote:

Behav & Hum Dec

Proc 272

(1996)

(attributing

ctions

against

self-interest

despite

full knowl-

edge

to visceral actors

defined

by

a

directhedonic

mpact

and an

effect on the relative

desirabil-

ity

of different

goods

and

actions).

2

Consider

Luigino

Bruni and

Robert

Sugden,

The Road Not

Taken:Two

Debatesabout

the Role

of

Psychology

n

Economics

1,

2

(unpublished

manuscript

002)

(on

file

with

author)

(explaining

hat the behavioral

conomicsmovement s

trying

o reverse

the

approach

nitiated

by

Paretoand others o

remove

psychology

rom

economics).

3

Later,

Milton

Friedman's

essay,

The

Methodologyof

Positive

Economics,

n

Milton

Friedman,

Essays

n Positive

Economics3

(Chicago

1953),

iberatedeconomists rom

even hav-

ing

to start with

empirical

egularities.

He

suggested

hat

simple

assumptions

with no

empirical

backingcould also serve as a startingpoint,because

they might

lead to accurate

predictions

despite being "wrong."

His "F-twist"s

best seen as

encouraging

reduced-form

tyle

of model-

ing

in

whichincorrect oundational

assumptionsmight

lead to

good

predictions

because

they

approximate

redictions

rom moreaccurate oundational

ssumptions

hat are left

unspecified.

4 W.

Stanley

Jevons,

The

Theoryof

Political

Economy

102-03

(Penguin

1970)

(arguing

that

knowledge

of the conditionsof

human

utility

are

necessary

o understand

conomics).



4/25


5/25

90

The

University

f Chicago

Law Review

[73:87

was

an

agreement

on

a convention for

how to do

economics,

ike

the

rules

of

tennis,

or

assuming

away

friction

n

physics.

Of

course,

the

equation

of

utility

and choice was also

historically

justifiable

in

the

early

1900s because

measuringutility directly,

at

that

time,

was as far-fetched

as,

say,

sending people

to the moon or

cloning

animals. n Pareto's

day

there was no

ability

to

place

electrodes

into a

monkey

or

human

brain

and

record

firing

rates

of

individual

neurons,

to alter

genes

in

mice and see what

those mice can't

do,

to record

online brain

activity

with 3-mm

spatial

resolution

using

PET or

fMRI,

to

identify

areas of brain

damage (lesions)

in

humans

and

see

what

skills those humans

ack,

to

genotype people

and

correlate

genes

with

behavior

and

skill,

to

give

humans

drugs

and

see

how

their behavior

changes,

or to create

"temporary

esions"

using

transcranial

magnetic

stimulation

TMS).

The

huge

advances

n all

these tools force

the readerto draw one

of two

conclusions:

1)

Pareto was

right

in his

own time-he and

his

peers

shouldn't

have

tried

to

measure

utility directly,

t was

hopeless-

and

his

intellectualheirs are still

right today,

because

their

tools

can't

possibly

measure

utility,

or

(2)

Pareto was

right

then because the tools

didn't

exist,

but the

neo-Paretians

are

wrong

now

because we have

tools to measureutilitythat didn'texist in Pareto'stime (andthe tools

will

only get better).

Conclusion

(1) might

be

right,

but

betting

that

something

is

impossible

in

science is

dangerous

and

neglects

the

op-

tion value of

exploring

wild ideas.

Conclusion

(2)

is the

progressive

conclusion

with

upside

potential.

The

ideas n this

Essay

are

not an insistence hat

revealed-preference

theory

has not been useful.

t

has,

and will continue o be. But the useful-

ness

of one

theory

ells us

nothing

aboutwhetheran alternative

pproach

mightprove equally

useful.

B.

Beyond

Revealed Preference:

Types

of

Utility

The

distinction between

utility

and

choice

has

been revived

in

two

ways.

First, Kahneman,

Wakker,

and

Sarin

noted that

one

can

concep-

tually

distinguish

our

types

of

utility:experienced

utility

(the

hedonic

sensation at the

time of

consumption

that

Jeremy

Bentham had in

mind);

remembered

utility;

forecasted

utility

(a

forecast of

experi-

enced

utility);

and

finally,

he familiarnotion of

decision

utility (num-

bersanobservercould use to rankanagent'srevealedpreferences).8

8

See Daniel

Kahneman,

Peter

P.

Wakker,

nd Rakesh

Sarin,

Back to Bentham?

Explora-

tions

of Experienced

Utility,

12

Q

J Econ

375,376-77

(1997);

Daniel

Kahneman,

New

Challenges



6/25

2006]

Neuroscience

nd

Paternalism

91

Thinking

ike a

psychologist,

t

is

easy

to

see

why

these

types

of

utility might

differ.

Remembered and

experienced

utility

can differ

because of

nostalgic

biases

in

memory (for example,

the

inability

to

recreate the

pain

of childbirth

n

imagination).

Forecasted

utility

could

be

subject

to

"projection

iases"relativeto

experienced

utility-people

buy

more

groceries

when

they're

hungry,'

nd

overestimate

how

badly

they

will feel

if

their favored candidate

loses an

election.10Most fun-

damental for

my

purposes,

decision

utility

could deviate

from

experi-

enced

utility

f

people

have not

yet

learnedwhat

they

like

(for

example,

children),

or

if

there is some

dissociationbetween

wanting

and

liking.

The

revealed-preferences

iew

basicallyequates

welfare

(experi-

enced utility)with choice (decisionutility),as a matterof the defini-

tion of welfare.Is this the best we can

do?

If

my

neighbor thumps

his

head

repeatedly

with a

ball-peen

hammer,

do I

have no

alternative

but

to infer that

hammering

his head

with a

ball-peen

hammer s the most

fun he can have?

A

more

sensible direction is to

study

the relation

between

experienced

and decision

utility empirically.

The

second revivalof the debate about

utility

is

work

by

the

psy-

chologist

Kent

Berridge

and

colleagues

on

"wanting,"

learning,"

nd

"liking"

hat addresses

the

measurement

challenge,

suggesting

some

waysin whichexperiencedutility(liking)and decisionutility (wanting)

might

differ,

depending

on

learning

hat

potentially

inks the

two."

My suggestion

in this short

Essay

is that the

wanting-learning-

liking

distinction

provides

a

potential

basis,

grounded

n

neuroscience,

for

asserting

hat choices are not

alwaysutility-maximizing.

n

fact,

the

learning part

of this frameworkwas

clearly

endorsed

by

Pareto as a

justification

for

assuming

that the

"subjective

act" of

utility

and the

"objective

act" of revealed

preference

are the same.

A

longer quota-

tion

containing

he short

excerptquoted

above,

reads:

We will studythe many logical,repeatedactionswhich men per-

form to

procure

the

things

which

satisfy

their tastes....

[W]e

are

concerned

only

with certainrelationsbetween

objective

facts and

subjective

facts,

principally

he

tastes

of men.

Moreover,

we will

simplify

the

problem

still more

by

assuming

that the

subjective

to

the

Rationality

Assumption,

50 J Institutional&

TheoreticalEcon

18,

21

(1994)

(discussing

the distinctionbetween

he two

primary

otionsof

utility: xperienced

and

decision).

9

See

George

Loewenstein,

Ted

O'Donoghue,

and Matthew

Rabin,

Projection

Bias in

Predicting

uture

Utility,

18

Q

J Econ

1209,

1215

(2003).

10 See DanielT.

Gilbert,

et

al,

Immune

Neglect:

A Source

of

Durability

Bias in

Affective

Forecasting,

5 J

Personality

& Soc

Psych

617,625

(1998).

11

See Kent

C.

Berridge

and

Terry

E.

Robinson,

Parsing

Reward,

6

TRENDS in Neurosci-

ences

507,

510-12

(2003)

(discussing

he interactionsof

wanting, earning,

and

liking

in condi-

tioned stimuli

as motivational

magnets,cue-triggered

"wanting," esponse

reinforcement,

nd

cognitive

ncentives).



7/25

92

The

Universityf Chicago

LawReview

[73:87

fact conforms

perfectly

to the

objective

fact.This can be done

be-

cause

we will

consider

only

repeated

actions to be a basis

for

claiming

hatthere s a

logical

connection

uniting

uch

actions.1

Pareto

clearly implicates learning

as a mechanism that

brings

wanting

and

liking together:

A

man who

buys

a

certain food for the first time

may buy

more

of it than is

necessary

to

satisfy

his

tastes,

price

taken into

ac-

count. But

in

a

second

purchase

he

will

correct his

error,

n

part

at

least,

and

thus,

ittle

by

little,

will

end

up by procuring

exactly

what he needs. Wewill

examine his action at the time when he

has

reachedthis

state.

Similarly,

f

at first he makes a mistake in his

reasoning

about what he

desires,

he will

rectify

t in

repeating

the

reasoning

and will

end

up by making

t

completely

ogical."

Thus,

the standard

assumption,

hat choices reveal "true"

utilities,

is

clearly

endorsed

by

Pareto

only

for choices

that

are

repeated

often

enough

to ensure that

learning

has occurred.The

wanting-learning-

liking

frameworkcan therefore be

seen as a

proper

generalization

of

revealed-preference heory

and,

in

fact,

it is

perhaps

the kind

of

gen-

eral

theory

Pareto

had

in

mind

in

the first

place.

Bruni and

Sugden

note that the "discovered referencehypothesis"s a revivalof Pareto's

idea.14

They

assert that

rationalchoice

theory applies

after

preferences

have been

"discovered"

y

trial-and-error

earning

and reflection.

But

this leaves a

large

and

important

hole:

what

happens

before

earning?

Besides

generalizing

revealed-preference theory,

the

wanting-

learning-liking

ramework

provides

a

potential way

to talk

scientifi-

cally

about how

restricting

or

imposing

choices

(paternalistically)

could

improve

welfare.

The core idea is

simple:

f

there are

separate

systems

for

recording iking,

expressing

wanting,

and for

learning

to

want what the brain likes, then paternalismcould be justified if the

wanting system

produces

choices that are

not later

liked,

and if

a

pa-

ternalistic

correction

produces

choices that

are unwanted

by

an

agent

but will

be liked

by

her,

or that are

wanted but

not

liked,

and if

the

correction

does not

cause other

harms

(or

much

harm to

rational

agents).

12

Pareto,

Manual

of

Political

Economy

at 103

(cited

n

note

6)

(emphasis

added).

13

Id

(emphasisadded).

14 See Bruniand

Sugden,

TheRoad Not Taken t7

(cited

in note

2) (labeling

he claim hat

rational-choice

heory

can

predict

behaviorwhencertain

criteriaare satisfiedas

the "discovered

preference

hypothesis").

ee

also CharlesR.

Plott,

Rational ndividual

Behaviour n

Markets nd

Social ChoiceProcesses:The

Discovered

PreferenceHypothesis,

n

KennethJ.

Arrow,

et

al,

eds,

The

RationalFoundations

f

EconomicBehavior

225,

226

(St.

Martin's

996)

(suggesting

a the-

ory

that

rational hoicesevolve

through

hree

stagesreflecting

xperience

and

practice).



8/25

2006]

Neuroscience nd

Paternalism

93

By

the

way,

note that

I

am

truly

not an

eager paternalist.

could

care less about

paternalism,

per

se.

Paternalism

s

only

interesting

as a

challenge

for behavioral

economics,

both

normatively

and

positively

(to

make sense of

paternalism

hat is

currentlypracticed,

and

changes

in

paternalism

across

time

and

geography).

At this

point,

the neuroscientificbasis

for

any practical

paternal-

ism

is

flimsy.

But

progress

n

neuroscience s

very

rapid.

And of

course,

there

is

a

genuine danger

of a

slippery

slope

in

moving

from

paternal-

ism that is

designed

to

be

unobjectionable,

because it

helps

some

and

doesn't

harm

others,

to

paternalism

hat

is

scary,

ntrusive and

wrong-

headed. The

political

economy

of

capture

by

special

interest

groups

also should make us wary, n practice, hat openingthe door to even

the mildest

paternalism

could

get

out of hand. But

it can't

hurt to

in-

stigate

a

purely

academicdebate about

how limited

paternalism

might

someday

be

scientifically

ustified, subject

to a reasonable

consensus

of

scientists,

awyers,

and citizens.

C.

Wanting-Learning-Liking

llustrated

I will motivate

my

discussion

by

starting

with details

of

a

simple

experiment

with mice

by

Pecifia and

colleagues."

Their

experiment

uses mice with a dopamine transporter DAT) "knockdown"due to

genetic engineering.

As

I

understand

t

(which

is not

well),

the knock-

down mice

produce

only

about 10

percent

as much

dopamine

trans-

porter

(which

"takes

up"

or

recycles dopamine)

as the control

mice,

but as

a

result

have 70

percent

more extracellular

dopamine sloshing

aroundbetween

neurons.

When

freely

allowedaccess o

food and

drink,

he DAT-knockdown

mice

consumed

twice as much as the control

mice.This

difference,

by

itself,

does not tell

you

whether

hey

liked the

consumption

r

just

wanted

it more.

The

mice were also

put through

a

standard

"runway"

ask,

in

which

they

are

kept hungry

for

eight

hours,

then trained

to move

down

a short

runway

from

a start

box to a

goal

box

containing

food.

During

a

learning

period,

the DAT-knockdown

mice ran about twice

as

quickly

down

the

runway

as the control

mice

and were much

less

likely

to

reverse course

or

pause

and

sniff.

Because

the knockdown

and

control

mice are about

equally

fast

in

a

pretraining

phase,

and

15

See Susana

Pecifia,

et

al,

Hyperdopaminergic

utant

Mice Have

Higher"Wanting"

ut

Not

"Liking"for

weet

Rewards,23

Neuroscience9395

(2003)

(finding

hat

chronically

levated

extracellular

opamine

n

mutant

micefacilitates

wanting"

utdoes

notincrease

"liking").



9/25

94

The

University

f

Chicago

Law Review

[73:87

equally

fast after a lot of

training,

his difference

during earningsug-

gests thatextra dopaminefacilitates

asterlearning.16

These

experiments

show that the DAT-knockdown mice

eat

more,

and learn there's food

waiting

in

the

goal

box faster than

the

control

mice

do,

but do

they actually

ike food and

liquid

more?

Here

we have

no

choice

but

to skate

out onto thin

empirical

ce: how do

you

measure

iking?

If

you

are

open-minded

about the

possible

difference

between

choice,

which is

observable,

and

liking

(true

utility),

which is not as

readily

observable,

hen

you

should

accept

that

any

measure

of

true

utility

won't

be as

sharp

and

agreeable

as

observing

choice is.

If

you

aren'twillingto acceptthis sadfact,stopreadingnow.

Pecifia and

colleagues

measure

iking

and

dislikingusing

physical

movements of the

mice--tongue

protrusions

and

paw licking

to

repre-

sent

liking,

and

gaping (mouth-opening,

ike a

yawn), screwing

up

the

face,

and

arm

flailing

to

represent

disliking.

Your first reaction

might

be

that these are not serious scientific measures.But

they

are taken

very seriouslyby people

who

study

animals,

because

they

are the best

availablemeasuresof

liking"

putting

observed

choice

aside,

of

course;

if

you

use choice to infer

liking

we're back

to

square

one

with

Pareto).

These expressionsand movements are also very similar acrossmice,

primates,

and human

infants;

n

fact,

they

are so similarthat

research-

ers use

an

equation (d

=

.26(w)32)

o

predict

the

duration

(d)

of

a fa-

cial

expression

(in

milliseconds)

n

terms

of a

species'

body

weight (w)

(in

kilograms).'8

In

any

case,

these

measures

of

liking

and

disliking

show

that the

DAT-knockdown

mice do

not

like

the sucrose solution

more

than

the

control mice.

In

fact,

1.0

M

sucrose,

the

largest

concentration,

s liked

significantly

ess

by

the

knockdown

mice."

16

See SamuelM.

McClure,

Nathaniel

D.

Daw,

and

P.

Read

Montague,

A

Computational

Substrate

or

Incentive

alience,

6 TRENDS n

Neurosciences

23,424

(2000).

17

Methods ike this

are used

by

pediatricians

o

judge

how

much

pain

children

are

in.

An

"oucher" cale is used

by

pediatricians

o

map

facial

expressions

nto a numerical

cale.

See,

for

example,

How to

Use the

Oucher,

online at

http://www.oucher.org/differences.htmlvisited

Jan

17,

2006).

See

also,

for

example,

Rita

Rubin,

Kids' Distress

s No Small

Thing,

USA

Today

D1-2

(May

9,

2005).

This can

be

thought

of as a

utility

unction hat

maps

facial

expressions

nto

nu-

merical

"true

disliking."

f the

goal

is

to

prescribe

an

optimal

dose of

painkiller, ou

can't

ust

let

a

child

"choose"a dose.

Therefore,

nferring

children's

disutility

pain--from

choice is

a bad

idea.

Inferring

heir

pain

from

facial

expressions

s a

good

idea.

18 See BerridgeandRobinson,26TRENDSinNeurosciences t 509

(cited

n note

11).

19

A

concern

in

the

study

is

that the

eating

task used

regular

food

(Purina

Rat

Chow)

and

water;

he

learning

ask

used Froot

Loops;

and the

liking

measuresused

sucrosesolution.I

pre-

sume

Pecifiaand

colleagues

are

comfortable

with the

working

hypothesis

hat

knockdown

mice

do not

have different astes for

any

of these

substances hancontrol mice do. See

also Kent

C.

Berridge

and

Terry

E.

Robinson,

What Is the Role

of

Dopamine

in Reward:

Hedonic

Impact,



10/25

2006]

Neuroscience nd Paternalism

95

This is

just

one

study

that

shows

a

potential

dissociation

between

wanting

and

learning,

which the

knockdown

mice

reveal

more

of,

and

liking,

which

they

reveal less of.

They

eat and drink

more,

but don't

seem to like it.

The same kinds

of

wanting-liking

issociationshave been

shown

in

experiments

with humans.20

Dopamine-receptorantagonists

appear

to

suppress

cigarette-smokerwanting

without

changing

(self-reported)

pleasure

rom

cigarettes.2'

Environmentalues create

earnedassociation

with

pleasurable

rug

use;

when

these cues are

presented hey

can

create

rapid craving,

ust

like the

conditionedstimulus

produced

salivation n

Pavlov's

dogs.

These

cues are

clearly

affecting

wanting, hough

they may

not affecteventual iking.

An

importantproperty

of the

wanting

and

liking systems

is that

they may

react

implicitly,

without consciousness.

For

example,

drug

addicts

will work

(lever-pressing)

or doses of stimulantsor

morphine

that are

so low that the addicts

report

no

subjective

effects

and have

no measurableautonomic

responses.

That

is,

the addictsdetect

enough

of the

drug

to

keep

wanting

the low

doses,

but don't

think

they

are

getting any dose at

all.22

Berridge

and Robinson

suggestgeneralregions

of the mouse brain

that areresponsible or wanting,iking,and

learning.23

Nobodyhas con-

structedsuch a

neurogeographical

map

of

utility

for the

human

brain,

and

it

will be

a

long

time before we have a

good

one.

(Furthermore,

utility

for

food,

shopping,

drugs,

ex,

and

golf may

all work

differently,

so we'll need a

book of brain

maps.)

But the

point

is that

such a

map

has

already

been

proposed

for

mice,

based

on a

relatively

recent and

modest amount

of research.So

it is conceivable that at least a

rough

Reward

Learning,

r Incentive

Salience?,

8

Brain

Rsrch Revs

309,

350

(1998) (finding

hat do-

pamine-depleted

atsstill

"like" ewards nd

simply

ail to "want"

ewards).

20

See

Berridge

and

Robinson,

26 TRENDS

in

Neurosciences

at 508

(cited

in

note

11)

(discussing

how

drug

addicts

will work for low doses of stimulants

r

morphinedespite

no sub-

jective

effect).

See also Carl

L.

Hart,

et

al,

Methamphetamineelf-administration

y

Humans,

157

Psychopharmacology

5,

80

(2001) ("[S]everal

nvestigators

have

reported

data

demonstrating

that

drug-related positive'

subjective

effects do

not

entirely

account

for

drug-taking

ehavior

andsome have

suggested

hat otherfactors hould

also be

considered.").

21

See Lisa

H.

Brauer,

t

al,

Haloperidol

Reduces

Smoking

of

Both

Nicotine-containing

nd

Denicotinized

Cigarettes,

59

Psychopharmacology

1,

34-35

(2001)

(discussingexperimental

resultswhere

haloperidol,

a

dopamine-receptor ntagonist,

was administered nd

significantly

reduced

moking

behaviorbut did not affect how much

he

subjects

iked

smoking).

22

See

Berridge

and

Robinson,

26 TRENDSinNeurosciences t508

(cited

n

note

11).

23

See id at

510

(containing

schematic

xplanation

f

regions

of the brain hatare linked

to

wanting, iking,

and

learning).

Their

analogous

ermsare

motivation

wanting)

andemotion or

affect

(liking).

dat 508.



11/25

96 The


Law Review

[73:87

understanding

f

wanting, earning,

and

liking

could

emerge

for

hu-

mans,probablyat first

for a

specialized

domain

ike

food.24

Having separate systems

for

wanting, earning,

and

liking

makes

sense

from an

evolutionarypoint

of

view.Take food as a

simple

case.

First

note that unless tastes for

specific

foods

are

truly

genetically

hardwired,

ome kind of

learning

system

is

necessary

to connect the

sensory

properties

of food-how it

smells,

feels,

and looks-to how

much an animal ikes

specific

foods.

This

basic

learning

s

what infants

do for

months-putting

everything

n

sight

in

their

mouths to see

how

things

taste.

An

infant who tastes dirt doesn't

anticipate

iking

dirt-

she

just

exhibits

wanting,

which is

really

just

wanton

exploration (to

create inputs for the learningsystem).Then she gradually earnsthe

difference

between

brown,

crumbly

stuff that tastes bad-dirt-and

brown,

crumbly

tuff that

tastes

good gooey

brownies.

The fact that the

learning system

exists

at

all is

prima

facie

evi-

dence

that

wanting

and

liking

are not

automatically

inked at

birth;

learning

s there for a

reason,

and the reason is to educate

the

wanting

system

about what is liked. But

why

would

wanting

and

liking

occur

in

separate

brain

regions?Berridge

and Robinson

propose

that

liking

for

food, sex,

and warmth

rely,

at

least to some

extent,

on

quite

different

sensory systemsfor processingsmell, heat,andtaste.A separatewant-

ing

system

is

needed to combine

these

signals

into a "common cur-

rency,"25

n

order

to

make tradeoffs between

goods

that are liked in

(sensorily)

different

ways.

Wanting

and

liking may

also

deviate,

usefully,

n

the

special

case

of

pregnancy.

When

my

wife was

pregnant,

he had a fierce

craving

or

lamb-one of her

favorite foods-then

almost vomited after

taking

one bite. Her

wanting

system

had

tuned to her own

tastes but the

cir-

cuitry

protecting

he delicate

fetus

registeredrapid

dislike.

Learning

s

the

mechanism hat trains

wanting

about what is

actu-

ally

liked

by

trial-and-error.

owever,

n

the

modem

economy,many

of

the

goods people

purchase

are

much

more

complex

o

process

and

rep-

24

A

similar

body

of research

n

humans ocuses attentionon

"actor-critic"

odelsof tem-

poral

difference

TD)

learning.

n

TD

learning,

n

organism

earnsa

value

function

according

o

an

adaptive

updating

equation

driven

by

the

prediction

error

(or

temporal

difference).

John

O'Doherty

and

colleagues

have

located areas of

striatum

in

the brain's

emporal

obe)

that

correspond

o "critic"

emporal

differences,

and

other areas that

correspond

o

an "actor's"

expected

reward

or forecast.

ohn

O'Doherty,

t

al,

DissociableRoles

of

Ventral nd

DorsalStria-

tum in

Instrumental

Conditioning,

04 Science

452

(2004).

In

terms of this

Essay,

the

actor is

wanting, nd he critic ystem s

learning,

which

adjusts

wanting

o it learns hetruevalue

(liking).

25

See Peter

Shizgal,

On the Neural

Computation

f Utility: mplicationsrom

Studies

of

Brain Stimulation

Reward,

n

Daniel

Kahneman,

Ed

Diener,

and Norbert

Schwarz, ds,

Well-

Being:

TheFoundations

f

Hedonic

Psychology

500,

509-10

(Russell

Sage 1999)(discussing

he

necessity

and existence of a

currency

unction to break down

different stimuli into decision

utility wanting)).



12/25

2006]

Neuroscience nd

Paternalism

97

resent than are

simple

foods. The

learning

necessary

to link

wanting

and

liking

of such

goods through

trial-and-error

einforcement s

ei-

ther

impossible (planning

the

perfect

once-in-a-lifetime

wedding)

or

very

slow.

Other mechanisms

may

substitutefor direct

learning

about

com-

plex

choices

(personal

advice, mitation,

advertising).

But it is

not

likely

that

the humanbrain has

developed

a

special

patch

for

unique impor-

tant decisions that

links imitation

directly

to actual

iking.

At

best,

hu-

mans

would

have a

capacity

to learn at a

young

age

whose

advice to

take

(i.e., they develop

a

tightly coupled wanting-likingoop

for

good

advice;

and advice

that is taken then

triggerswanting

or a

good,

which

is a sensible forecast

of

later

liking

because that's what

"good

advice"

means).

Such a

systemmight

work

well,

but

probably

not

perfectly,

and

might

be vulnerable

o

exploitation

by profit-maximizing

irms

(for

ex-

ample,celebrity

endorsements

mighthijack

uch a

system).

II. HUMAN MISFIRES

IN WANTING AND LIKING

A. Disorders

The

most

compelling examples

of

wanting-liking

aps

are

patho-

logical

disorders hat

people clearly

would

like

to

change

if

they

could.

Obsessive-compulsive

disorder

(OCD)

is a

psychiatric

disorder

characterized

by

obsessive

thoughts

and

compulsive

actions,

such as

cleaning,

checking,

counting,

or

hoarding.26

"Cleaners"

report

an

in-

tense

feeling

that

their hands are not

clean,

and

must

be washed over

and

over to reduce that

feeling. "Checking"

s

similar.

My

father used

to

compulsively

check the four oven stove dials

in

our house before

leaving

the house-he

would

point

to each

and

say

"off,

off, off,

off,"

partly

coping

with embarrassment

y making

fun

of his

mild

checking

compulsion.

Sometimes he would

stop

the

car,

halfway

out

the drive-

way,

and

get

out to double-check hat the frontdoorwas locked.

In

the

revealed-preferences pproach,

we

would

infer

that

compul-

sive cleaners

have a

high utility

for

clean hands.Some sensible

predic-

tions

may

come from this

inference--for

example,

cleaners

spend

a

lar-

ger

income share

on

cleaning

han noncleanersdo.

But the

revealed-preferences

view

only gets you

so far.

If

you

give

compulsive

cleaners

a serotonin

reuptake

inhibitor

(SSRI)

like

fluoxetine

(Prozac)

or

paroxetine

(Paxil),

a

majority

of cleanersfind

26

Note that the

DAT

knockdownmutant

mice exhibit

"sequential uper-stereotypy"

n

grooming

hat

is

remarkably

imilar o OCD

cleaning

n humans.See Kent C.

Berridge,

t

al,

Sequential uper-stereotypy

f

an Instinctive ixedAction Patternn

Hyper-dopaminergic

utant

Mice:

A

Model

of

Obsessive

Compulsive

Disorder and

Tourette's,

(4)

BMC

Biology (2005),

online at

http://www.biomedcentral.com/content/pdf/1741-7007-3-4.pdf

visited

Jan

17,

2006).



13/25

98

The

University

f

Chicago

Law

Review

[73:87

that

their

symptoms

are diminished.Of

course,

we can

easily

stretch

revealed-preference

anguage

to allow

"state-dependence"

f clean-

ing-utility,

which is altered

by

these

drugs.

But

patientsactively

seek

out these

drugs

to

change

their

preferences.

To talk about the demand

for

a

cure,

we need

to

posit

a

"metapreference"

n which cleaners

pre-

fer to be

symptom-free

han to be

compulsive.

Their

metapreference

s

expressed

n demand

for

Prozacor Paxil.

A

big

advantage

of

revealed-preference

heory

is

simplicity.

But

addingconcepts

of

state-dependence

nd

metapreference uickly

makes

the

theory

complicated,

not

simple.

And even

if enriched revealed-

preference

theory

can accommodate hese

phenomena,

he details of

how

preferences

are linked to

genes

and childhood

behavior,

how

"states"ike

SSRI

drugs

work,

and whether

agents

understandhe source

of

their behaviorand the

state-dependence,

which all matter for

good

prediction,

equire

a lot of

empirical nput

rom

outside

economics

per

se.

The

language

of

wanting

and

liking

could

be even

simpler.

In

wanting-liking

erms,

an

OCD

patient

wants

to clean but does not

like

it. Most

say

that

they

do not

like

cleaning

per

se

-

it

can be

very

time-

consuming,

doesn't

bring

pleasure,

and direct measurement

(facial

expressions,self-report, maging

of striatum

activity)

would

probably

show that

liking

is low. But such a

patient

is

motivated,compulsively,

to clean

(she

"wants"

o).

SSRI

drugs

reduce

wanting.

Patients

take

medication to override

their

wantingsystem.

The

OCD

example

is a

relatively easy

one. Of

course,

the

want-

ing-liking

framework s useful

only

if

there are more common behav-

iors,

with

substantialeconomic

consequences,

hat

can be seen

as re-

flecting wanting-liking aps.

Most of

the

obvious

examples

have to do

with time

preference,

which

I'll

discuss

n

the next Part.

Another

interesting example

is

the behavior

of

"shopaholics."

Compulsive hopping

s a disorder

with formal

diagnostic

riteria

analo-

gous

to those for

compulsivedrug

use

or

gambling.27

will use

it as

a

casual

example

just

because it

is

a name for a

possible

disconnectbe-

tween

the

transitory leasure

from

buying

goods

and the later

pleasure

from

consumption.

A

different class

of behavior nvolves

"virtuous

n-

tentions" ike

buying

exercise

equipment

or

a health club

plan."8

27

See

LorrinM.

Koran,

et

al,

Citalopram

Treatment

f

Compulsive

Shopping:

An

Open-

LabelStudy,63 J ClinPsychiatry 04,704 (2002)(showing hat treatmentwith the antidepres-

sant

citalopram

Celexa)

reduces

hopping ompulsion).

28

Stefano Della

Vigna

andUlrike

Malmendier,

aying

Not to

Go

to the

Gym

22

(unpub-

lished

manuscript

2005),

online at

http://emlab.berkeley.edu/users/sdellavi/wp/gymemp05-04-

20.pdf

(visited

Jan

17,

2006) (explaining

he irrationalbehaviorof

gym

users' contractchoices

based on theiroverestimation f future

efficiency

or

self-control).



14/25

2006]


99

An

economist's

definition

of

shopaholism

s

systematically

buying

goods that are not later

consumed.9

Such

a

disorder

could result

from

a

wantingsystembuying goods

that

the

likingsystem

doesn't

appreci-

ate

(like

closets

full of shoes that are never

worn).

Of

course,

in

the

language

of belief and

preference,

one could

argue

that

shopaholics

are mistaken

n their beliefs

about future

consumption or

their

budget

of time to use

goods).

Then

the

challenge

s

explainingwhy they

don't

learn

from

a chronic

pattern

of

purchase

and disuse.

An

extremeform of

shopaholism

s an

OCD disorder

called

"col-

lecting"

or

"hoarding"

n

which

people

collect items to an

extreme.

A

milder

orm

s

probablypresent

n all

of us:

most

people

have at least

one

class of

goods

or services

they enjoy shopping or,

but

they

don't have

enough

time or inclination o

actually

consume.Mine is

books-I love

getting

books,

but

never

get

around o

reading

hem all.

My

wife

is ob-

sessed

with

buying

marinades.

We

have around

wenty,

whichare invaria-

bly opened

once and then

sit

patiently

or

years, hree-quarters

ull.

B.

Wanting

and

Integration

of

MomentaryLiking

over Time

Food is a useful kind

of

good

for

exploring

the neural basis of

wanting

and

liking,

because mechanisms

are

common across

species

and consumptionis immediate. But for most goods and services of

economic

interest to humans

education,

savings,

durables,

redit card

use-costs and

benefits are

spread

out over

time. It

may

be

very

un-

comfortable

to learn

to windsurf or to

struggle

through college

(the

liking system

is

not

happy

during

the

learning),

but

rewards

come

in

the

form

of later

liking (windsurfing)

r future

iking

of

goods

that can

be

bought

with

increased

wages

from a

college

education.

The

wanting

system

has to

integrate

these

likes and dislikes over

time

to determine

overall

liking.

One method for integrating iking is exponentialdiscountingof

future

rewards,

which

corresponds

to

dynamic consistency

in choice

and has much normative

appeal.30

A more

general approach

s

models

of

present-biased

preferences

(for example,quasihyperbolic

iscount-

ing),"

which

splice

a

preference

for

immediacy

and conventional

dis-

29

The closest

clinicians ome to

an

economic

definition

s the

"[f]requent

uying

of

more

than can

be

afforded,

more

than is

needed,

or

for

longer periods

of time

than intended."

Kim

Bullockand

Lorrin

Koran,

Psychopharmacology

f CompulsiveBuying,

39

Drugs

of

Today

695,

696

(2003).

30 Forshort-lived

pisodes

where

discounting

s

likely

to be weak

(for example, valuating

an

operation,

movie or a weekend

getaway),

Kahneman,

Wakker,

nd Sarin

propose

and axio-

matize a rule that takes

the

integral

of

momentaryiking

over time.112

Q

J Econ at 390

(cited

n

note

8).

31

See Ted

O'Donoghue

and Matthew

Rabin,

Doing

It Now

or

Later,

89Am EconRev

103,

106

(1999) (discussingpreferences

as

time-inconsistent r

present-biased,

meaning

that a

per-



15/25

100

The

University

f Chicago

Law Review

[73:87

counting

of future

rewards.32

hese

preferences

can be characterized s

reflecting

behaviorof a

wanting

system

that

weights

current

iking

too

heavily,compared

to

exponentialdiscounting.

Let's

consider two ex-

amples:

addictive

drugs

andcredit card

spending.

1. Addictive

drugs.

The hallmarks

of

addictive

drugs

are tolerance

(larger

doses are

needed to

produce equal

hedonic

sensations)

and withdrawal

(un-

pleasant physical

sensations

during

abstinence).

A naturaland fruitful

way

to model these effects is

through state-dependence

of current

utility

on

past

consumption;"3

r a richer

view in whichstates are

nu-

mericalsummariesof use historiesthat go up or down when addicts

use

or

abstain.4

In

any

case,

a

central ssue

is

whether

potential

addicts

rationally anticipate

how current

consumption

creates

tolerance

and

possible

future withdrawal.

A

key prediction

of the

rational addiction

model is that

current

consumption

responds

to

expected

future

price

changes,

as it

appears

o do."

Aggregated

data are too

coarse,however,

to tell whether some

potential

addicts refuse to

start,

or whether ac-

tive

users

quit,

n the

face of future

price

increases.

In

the

wanting-liking

view,

addiction means that

wanting

can

be

createdby a desire to avoidthe painof withdrawal, r is triggeredby

craving

("self-medication"),

ut the

drug

administrations not

actually

son's

relative

preference

or

well-being

at

an

earlierdate versus

a

laterdate

will

changedepend-

ing

on when he

or she

is

asked);

David

Laibson,

Golden

Eggs

and

Hyperbolic

Discounting,

12

Q

J Econ

443,445-46

(1997)

(discussing

ow

decisionmakers

oresee their own

dynamically

ncon-

sistent discountrates over short

horizons

(high)

and

long

horizonsand commit

themselvesto

nonliquid

assetsto

constrain heir

future

elves).

32

See

SamuelM.

McClure,

t

al,

Separate

Neural

Systems

Value mmediate nd

Delayed

Monetary

Rewards,

06

Science

503,

506

(2004) (describing

how

separate

functions xist in the

brainassociatedwithimmediatepreferences possiblyreflectingevolutionary daptations)and

abstract uture

preferences

and

how

these

separate

unctions

may explain

the

idiosyncrasies

f

human

preferences).

33

See

generally

Gary

S.

Becker and KevinM.

Murphy,

Theory f

Rational

Addiction,

6

J Polit Econ 675

(1988)

(describing theory

of

rational

addiction hat

is

partially

based on

what

state of addiction n

addict

s

in).

34

See Bernheim

and

Rangel,

94 Am

Econ

Rev at

1565

(cited

in

note

1)

(modeling

addic-

tive states

by

incrementing

r

decrementing

he

state

of

addiction

based

on

decisions o use

or

not

to use-a

higher

state

corresponds

o

greater

addiction,

or

example,

omeone in state 8 is

more addicted han

someone n

state

5).

35

See

Gary

S.

Becker,

Michael

Grossman,

nd Kevin M.

Murphy,

n

Empirical

Analysisof

Cigarette

Addiction,

84 Am

Econ Rev

396,

397

(1994)

(discussing

how a

one-period

price

in-

crease of 10

percent

reduces

consumption

n the

previousperiod by

0.6

percent).

See also Jona-

than Gruber

and Botond

Kbszegi,

s Addiction"Rational"?

heory

and

Evidence,

116

Q

J Econ

1261,

1273

(2001)

(showing

hat consumers

ncrease

consumption y

hoarding

cigarettes

n an-

ticipation

of futuretax

increases);

Angela

A.

Hung,

A Theoretical nd

Empirical tudy

ofAddic-

tion

86-87,

unpublished

hD

dissertation,

California nstituteof

Technology2001)(finding

hat

cigarette

consumers re

forward-looking

n

respect

o

prices).



16/25

2006]


101

pleasurable.

(At

best,

it cures the

withdrawal

symptoms;

but other

cures,

like

rehabilitation

or

methadone,

which do

not feed

the addic-

tion as

self-medicating

doses

do, may

be

available.)

As

Bernheim and

Rangel suggest,

it is sensible to

entertain

the idea

that when

craving,

addicts'

wanting

systems

are not

fully

accounting

or

the

future dislike

(or internality)

created

by

currentuse. This

phenomenon

is

not

easily

explained by hyperbolic discounting,

because

exogenous

cues--like

seeing drug

use

in

a

film--can

triggercraving.36

2. Creditcard

spending.

For the sake of

argument,

uppose

that at least some

consumers

chargetoo much on their creditcards, n the sense of revealinga want-

ing-likinggap (at

least

ex

post).

This

may

be due to

poor

understand-

ing

of how interest

rates

work

(for

example,

how

rapidly

nterest com-

pounds)

or insufficient

appreciation

or the status

quo

bias of

sticking

with one card after an introductory teaserrate"period lapses."

Another

possibility,

n

wanting-liking

erms,

s

that the brain likes

walking

out

of a store

(or

clicking

an internet radio

button,

or

calling

an 800-numberand

imagininggoods being shipped),

but the

wanting

system

does not

anticipate

the

"pain

of

paying"

when

the

credit card

statement arrives."That is, currentwantingdoes not integratefuture

dislike.

III.

PATERNALISM

I

adopt

the textbook

definition

of

paternalism

as "the interfer-

ence of a state or an

individual

with

another

person, against

his

will,

and

justified

by

a claim

that

the

person

interfered

with will

be better

off or

protected

from harm."39

In

my

terms,

paternalism

could be

justified,

n terms

of

a

person'sown

welfare,

f the

wanting system

does not

produce

what the

liking

36

See David

Laibson,

A

Cue-

Theoryof

Consumption,

16

Q

J Econ

81,

82-83

(2001) (ex-

plaining

a cues model that assumesthat the

presence

of cues

(sensory inputs)

linked to

past

consumption

ffectshabit

ormation).

37 See

Haiyan

Shui and

Lawrence

M.

Ausubel,

Time

nconsistency

n

the CreditCardMar-

ket

1,

27-28

(14th

Annual

Utah

Winter

Finance

Conference,

May 2004),

online at

http://

papers.ssrn.com/sol3/papers?.cfm?abstract_id=586622

visited

Jan

17,

2006)

(explaining

hat a

hyperbolic

model

s

necessary

o

explain

consumerbehavior

n

the creditcard

market).

38

See DrazenPrelec and

George

Loewenstein,

TheRed

and

the

Black:Mental

Accounting

of Savings

and

Debt,

17

Marketing

Sci

4,

23

(1998) (describing

mental methods

includingpro-

spective

accounting

nd

decoupling

hatallow consumers

o

enjoy acquisitions

t the moment

of

purchase

with

a

credit

card without

considering

he

unpleasantness

f

paying

their

credit

card

bill when t comes

due).

39

Gerald

Dworkin,

Paternalism,

dward

N.

Zalta,

ed,

The

StanfordEncyclopedia

f

Phi-

losophy 2006),

online at

http://plato.stanford.edu/entries/paternalism

visited

Jan

17,

2006).



17/25

102

The

University

f Chicago

Law Review

[73:87

system

likes

and

if

the

intervention

creates

more

liking

than the

per-

son would achieve on his own or with market-supplied

elp.40

Ideally,

such

paternalism

would

be "libertarian"n the sense

that

it would

only

intervene

n

choice

when

some

kind of

regulation

s

nec-

essary

(for

example,setting

a

default),4'

nd

would be

"asymmetric"

n

the sense

that little harm would be done to

people

whose

wanting

and

liking systems arewell-synched.42

Paternalism ould take

two

mild

forms,

icensing

and

dramatizing.

"Licensing"

s

checking

whether

wanting

and

liking

are

sufficiently yn-

chronized o

permit

a

person

to make choices on his own

(presumably

in a

particular

domain).

"Dramatizing"

s

using

mechanisms

o remind

the

wanting ystem,

at the time of

choice,

about

iking

that

may

be

over-

looked or

missynchronized

ith

wanting.

A.

Licensing

Driving

tests

provide

a

model for how

licensingmight

occur,

as a

way

to

certify

sufficient education of the

wanting

system.

Because

adolescents

develop

the

hand-eye

coordinationand

patience

to drive

safely

at different

ages,

and

learning

to drive takes time

and

supervi-

sion,

the current

policy

makes sense:

prohibit driving

by

all

children

up to a certain

age;43

issue learner'spermitsthat allow driving only

when

a

supervising

adult is

present

(so

the learners can

learn);

then

force

would-be drivers

to

pass

written and

physical

driving

tests,

pref-

erably

with

periodic

retesting

and

other constraints

(such

as license

suspension

after

accidents).

One can

imagine

a "credit-card

ermit"

test that

is

conceptually

similar o a

driving

est.Treata

creditcardas a

piece

of

equipment

hat

40

A

shadow

s

always

cast

on

this

type

of

analysisby

the

naggingquestion,

"How

can we

be so sure thattheywon't ike it?"Thereare twocentralproblems: slippery lope,andcapture.

The

slippery

lope

argument

s

that

interfering

with

OCD

patients

might

make

it

legally

or

po-

litically

easier

for

more

interference

n

similar

behaviors hat are less

scientifically rounded

as

disorders.The

capture

concern

s that

paternalistic

policies

will be

guided

or

shaped

(in

their

detail)

by

special

interestswho

either

profit

from

interference

for example,

prohibitingprice

advertising

or

eyeglasses)

or

who want to

"saveothers"

for

example, eligious anatics).

A

full

analysis

of

any

policy

should,

of

course,

anticipate

both

slippery lope

escalationand

capture.

t is

too muchto ask

at this

early

stage,

however,

when

detailsof even a

singlepolicy

are

speculative,

to

anticipate

hese

problems

and

respond

o

them,

at least

in

this short

Essay.

41

See

generally

RichardH.

Thalerand

Cass

R.

Sunstein,

Libertarian

aternalism,

3

Am

Econ Rev

175

(2003);

Cass

R.

Sunsteinand RichardH.

Thaler,

Libertarian

aternalisms Not an

Oxymoron,

0 U Chi L Rev

1159

(2003).

42

See

generally

Colin

Camerer,

t

al,

Regulation

or

Conservatives:ehavioralEconomics

and the Case

or "Asymmetric

aternalism,"

51 U Pa L Rev 1211

(2003).

43

Some states have

a "farm icense"that allows

very young

children o drive

in

certain

situations.For

example,

n

North

Dakota,

fourteen-year-olds

an

drive farmvehicleswithin 150

miles of

their farm.License

Classifications

North

Dakota

Dept

of

Transportation005),

online

at

http://www.state.nd.us/dot/dlclass.htmlvisited

Jan

17,

2006).



18/25

2006]

Neuroscience nd

Paternalism

103

a user should

know how to

operate.

Require potential

borrowers to

take a

quiz

with

some

basic

questions,

ike

how

rapidly

nterest

com-

pounds

and how

long

it

takes to

pay

off a

$1,000

balance

if

minimum

payments

are made

and

interestaccumulates.

erhaps

have

them

speak

with

a

counselor

or with

experienced

adults,

ome of whom

used cards

wisely

and

others

who

declared

bankruptcy

nd

regret

t. The test could

also use

the

opportunity

o educate

potential

borrowers bout

basicbor-

rowing

rates

and

phenomena

(much

as

driving

ests remind

people

of

traffic

aws and some

facts).

The

paternalism

ere

is

asymmetric

ecause

if

borrowers

now these

facts,

he test can be

designed

o be

easy

to

pass.

B.

Dramatizing

Another

mild

paternalistic

ntervention could be

used

in

cases

where one

suspects

the

wanting system

is

not

accounting

for

future

liking

or

disliking

correctly,

by bringing

uture

liking

and

disliking

viv-

idly

into

the

present.

This is often done in "scared

straight"

ilms

and

talk-show

nterventions hat show

juvenile delinquents

what

prison

is

like.44 lcoholics

Anonymous

(AA) (one

of

the more successfultreat-

ment

programs)

uses vivid testimonials

by

other

AA

members that

remind

sober alcoholics

of

all the

crazy

and destructive

things

that

alcoholics do when they are drinking. Teenagers in high school

programs

can be made to

carry

around

a

baby

who

periodically

simu-

lates an

infant's

typical

needs

(crying,changing,nursing),

o

dramatize

the constraint hat results

from

early parenthood.

A

combination

of

nutritional

modeling

and

computer morphing

of a

body image

could

be used to show a

person,

for

example,

what

they

would look like

in one

year

if

they

continue

their

steady

diet

of

fast

food

or,

oppositely,

f

they

stuck with their

personal

trainer three

times a week for a

year.

C. Calibration

A

strong paternalistic

ntervention

s to

try

to calibrate

iking

di-

rectly,

and then

require

choosers to consider

their

liking ratings

when

making

choices

(weak form)

or

to force

them to choose the

goods

they

liked

most

(strong

form).

Of

course,

the success

of this

technique

depends

on how well

liking

s measured.

44

Unfortunately,

he available vidence

suggests

hese

programs

on'twork

very

well. See

Surgeon

General,

YouthViolence:

A

Report of

the

Surgeon

General

95

(2001),

online at

http://media.shs.net/ken/pdf/surgeon/SG.pdf

visited

Jan

17,

2006)

(discussing

esults that show

shock

programs

uch as Scared

Straight

have

equal

or

higher

rearrest ates than

exposure

o

no

program

t

all).



19/25

104 The


Law

Review

[73:87

The

kind of

thing

I

have

in

mind is a blind

taste

test-tasters

rate

objects,

and

then

automatically

hoose the one

they

rated most

highly.

This could

improve

choice if

there is

a

component

of

wanting,

ike

a

brand

name,

that is not a

genuine component

of

liking.

But of

course,

if

taste

is not the

only

component

of

liking,

then

true

liking

is mis-

measured

by

this

kind of calibration.

D.

Delegating

Choice

When

there are

wanting-liking

aps,

it

is

possible

that

somebody

else knows

more

about what

you

would

really

like

than

you

do.

Hav-

ing

the local

wine

connoisseurat

dinner

choose

the wine

for

the

table

is a common and good practice.From our point of view,delegating

choice is wise

if

the

expert

knows about

typicalpatterns

of

mistakes--

commonly

wanted choices that

are

typically

disliked later

("don't

fill

up

on

bread;

save room for

dessert").

Allowing experts

some

say

in

a

person's

choice

is

a form

of

paternalism;

n

the weak

form,

the chooser

is forced to hear an

expert opinion

(for

example,

a

marriage

coun-

selor's

report);

n the

strong

form,

the

expert

can overrule

the

choice

or

impose

a

choice.4"

Arranged

marriages

re an

interestingexample

of

various

degrees

of this sort of delegation.Ignoringexternalities ike familypride and

business

connectionscemented

by

marriage,

he

argument

or

arranging

a

marriage

s that

members of a

prospective

couple

know less about

what kind of

people they

will

truly

like

in the

long

run than a match-

maker

(often

a

parent)

does. The

couple's

judgments might

naturally

overweigh

observableshort-run

attributes

for

example,

attractiveness

and

sex

appeal)

and

underweigh

ess observable

ong-run

attributes

for

example, idelity

or

good

parenting),

while wiser

and older matchmak-

ers

weight

these

attributesmore

accurately.

Indeed,a smallpartof thisconcept s alreadybuilt ntolaw: n all but

three

American

states,

minimum

marriageages

are

lower with

parental

consent than

without

consent.This

gap

in

consent

ages

can be seen as

a

policy

that

permits

marriage

at

age

fifteen

(in

Utah,

for

example),

but

with

a

parental

eto

thatcan

delay

marriage

ntil

age

eighteen.46

E.

Promoting

Learning

In

the

liking and learning. neuroscience and paternalism - c. f. camerer

Documents