Transcript here`s - Association for Contextual Behavioral Science

Gregory J. Madden
Utah State University
Acknowledgements



NIH: RO1 DA 029100, RO1 DA 029605
USDA: 59-5000-1-0033, 59-5000-0-0065
Faculty Collaborators:





Amy Odum
Tim Shahan
Michael Twohig
Heidi Wengreen
Graduate Students Collaborators:

Current



Jay Hinnenkamp
C. Renee Renda
Kate Morrison
Jillian Rung
Past


Patrick Johnson
Rochelle Smits
Brooke Jones
Jeff Stein
Gregory J. Madden
Utah State University
Behavioral Economics
Dan Ariely
Kahneman &
Tversky
Thaler &
Sunstein
Behavioral Economics
George Collier
S. R. Hursh
BEHAVIORAL Economics
Behavioral ECONOMICS
Behavioral Economics
Leonard Green
George Ainslie
Warren Bickel
BEHAVIORAL Economics
Howard Rachlin
Behavioral ECONOMICS
Gregory J. Madden
Utah State University
Varieties of Impulsivity
(Evenden, 1999)

Response inhibition failures
 Blurting out, can’t resist the urge to eat/drink

Rapid internal clock
 It seems to take forever for the light to turn green.

Acting without thinking
 Failure to consider possible outcomes

Devaluing future outcomes
 Future consequences weigh little in decision-making
 Delay discounting
Delay Discounting: What is it?

As the interval between response and
reinforcer increases, the value of the
reinforcer declines.
Value
Delay Discounting: What is it?

As the interval between response and
reinforcer increases, the value of the
reinforcer declines.
Value
Delay
Delay Discounting: What is it?

As the interval between response and
reinforcer increases, the value of the
reinforcer declines.
Value
Delay
Delay Discounting: What is it?

As the interval between response and
reinforcer increases, the value of the
reinforcer declines.
Value
Delay
Delay Discounting: What is it?

As the interval between response and
reinforcer increases, the value of the
reinforcer declines.
Value
Delay
Delay Discounting
As the interval between response and
reinforcer increases, the value of the
reinforcer declines.
Richards
et al.
(1997)
Value

Delay
Determining discounted value…
Determining discounted value…
Determining discounted value…
They both have
the same value
to me.
I’m indifferent
% Subjective Value
of Larger-Later Reward
100
75
50
25
0
0
10
20
Delay
30
30 sec delay
30 sec delay
I’m indifferent
30 sec delay
% Subjective Value
of Larger-Later Reward
100
75
50
25
0
0
10
20
Delay
30
Value
% Subjective Value
of Larger-Later Reward
100
Richards
et al.
(1997)
75
Delay
50
25
0
0
10
20
Delay
30
% Subjective Value
of Larger-Later Reward
100
A
Vd = -----------(1 + kD)
75
50
25
0
0
10
20
Delay
30
Delay Discounting

What is it?
 As the interval between response and
reinforcer increases, the value of the
reinforcer declines.

Why discount the future?
On the Origin of Discounting the
Future

If I eat the smallerreward,
Butsooner
if I choose,
the
then
my chances
larger-later
my
of chances
survivingof(and
mating)
are are
slightly
surviving
improved.
greatly
improved.
Let’s do
it! not
These genes
will
A foraging animal…
pass to the next
generation.
Smaller-sooner
reward
Larger-later
reward
Stevens & Stephens (2010)
On the Origin of Discounting the
Future

This rat starves and
A foraging animal…
does not pass its
genes on to its pups.
Stevens & Stephens (2010)
Delay Discounting

What is it?
 As the interval between response and
reinforcer increases, the value of the
reinforcer declines.

Why discount the future?
 Why discounting is important.
% Subjective Value
of Larger-Later Reward
Why is it important?
100
The shape of the discounting
curve predicts some
interesting instances of
irrational choice.
75
50
25
0
0
10
20
Delay
30
Value
100
75
50
25
0
400
300
200
100
Time to Reward Delivery
0
Value
100
75
50
25
0
400
300
200
100
T1
0
Value
100
75
50
25
0
400
300
200
100
T1
0
I REALLY
love
hamburgers!
Value
100
75
50
25
0
400
300
200
100
T1
0
Rational Choice
I REALLY
love
hamburgers!
100
75
50
25
0
400
T2
300
200
100
T1
0
Irrational Choice
100
75
50
25
0
400
T2
300
200
100
T1
0
Irrational Choice
I REALLY
hate myself!
100
75
50
25
0
400
T2
300
200
100
T1
0
Rational Choice
Value
100
75
50
25
0
400
300
200
100
T1
0
Rational Choice
Value
100
75
50
25
0
400
T2
300
200
100
T1
0
Gregory J. Madden
Utah State University
Neutral
Bad
Really Bad
Extreme
Suffering
Madden, Petry, Badger & Bickel (1997)
Control
Opioid dependent
100
Value
80
60
40
20
0
0
2500
5000
Delay (days)
7500
10000
Bickel, Odum, & Madden (1999)
Control
Current Smokers
100
Value
80
60
40
20
0
0
2500
5000
Delay (days)
7500
10000
Extensions

Alcohol
 Vuchinich & Simpson (1998)

Methamphetamine
 Hoffman et al. (2006)

Cocaine
Articles Published per Year
(source: PubMed)
Key Word: Delay Discounting
250
200
150
100
50
0
1980
1990
Publication Year
 Coffey et al. (2003); Heil et al. (2006)

HIV-risk behaviors
 Odum et al. (2000); Chesson et al. (2006)

2000
Meta-analysis (MacKillop et al., 2011)
 Fail-safe N > 4,500 unpublished studies
2010
Delay Discounting



What is it?
Why discount the future?
Why is it important?
 Part 1: Shape of the discounting function
predicts irrational choice.
 Part 2: Individuals differ in the degree to
which they devalue future consequences
and this correlates with addiction.
 But correlation is not causation
Which Came First?
(Carroll & Perry, 2008; Stein & Madden, 2013)
Acute or chronic use of
drug produces neuroadaptations leading to
steep delay discounting.
 Steep discounting puts
the individual at risk of
drug dependence.
 3rd variable

Drug Effects on Delay Discounting

de Wit & Mitchell (2010)
 Consistent effects of acute drug
administration on delay discounting are rare
and, when observed, there are very few
studies.

Stein & Madden (2013)
 Chronic drug administration likewise does
not produce consistent effects on delay
discounting (see also Setlow et al., 2009).
Bickel, Odum, & Madden (1999)
Control
Current Smokers
Ex-Smokers
100
Value
80
60
40
20
0
0
2500
5000
Delay (days)
7500
10000
Bickel, Odum, & Madden (1999)
100
Value
80
Likely Quitters?
60
Unlikely Quitters?
40
20
0
0
2000
4000
6000
Delay
8000
10000
Does Steep Discounting Precede
& Predict Drug Taking & Abuse?

Self-Report Data:
 Kollins (2003): steep discounting is
significantly correlated with…
○ Age of first cigarette (r = -.51)
○ Age of first alcoholic beverage (r = -.34)
○ Age of first marijuana use (r = -.48)
○ Number of different drugs used (r = .32)
○ Number of times passed out from intoxication
(r = .73)
Does Steep Discounting Precede
& Predict Drug Taking & Abuse?

Longitudinal Data:
 Steep delay discounting predicts adolescent
alcohol, cigarette, and other drug use
(Audrain-McGovern et al., 2009; Brody et
al., 2014; Khurana, et al., 2013; Kim-Spoon
et al., 2014)
Does Steep Discounting Precede
& Predict Drug Taking & Abuse?
Discounting appears to predict treatment
success:
100
 Dallery & Raiff (2007)
 Krishnan-Sarin et al. (2007)
 MacKillop & Kahler (2009)
 Washio et al. (2011)
80
Value

Likely Quitters?
60
Unlikely Quitters?
40
20
0
0
2000
4000
6000
Delay
 Yoon et al. (2007)
 Although see Landes et al. (2011) for a
published exception
8000
10000
Does Steep Discounting Precede
& Predict Drug Taking & Abuse?

Animal studies…
Animal Longitudinal Studies

Research strategy...
Steep
Discounters
Shallow
Discounters
Perry et al. (2005)
Hi Imp
Low Imp
% Subjects Meeting
Acquisition Criterion
100
50
0
0
5
10
15
Session
20
25
30
Does Steep Discounting Precede
& Predict Drug Taking & Abuse?

Acquisition
YES
NO
Cocaine
Anker et al. (2009)[p = .08]
Kosten et al. (1997)
Broos et al. (2012)
Perry et al. (2005)
Perry et al. (2008, Exp. 2)
Perry et al. (2008, Exp. 3)
CPP Studies:
Kosten et al. (1994)
Yates et al. (2011)
Does Steep Discounting Precede
& Predict Drug Taking & Abuse?

Acquisition
YES
NO
Cocaine
Time Spent in Amphetamine Place
(difference from baseline)
Anker et al. (2009)[p = .08]
Kosten et al. (1997)
Broos et al. (2012)
Perry et al. (2005)
Yates et al. (2011)
Perry et al. (2008, Exp. 2)
400
HiI
Perry et al. (2008, Exp. 3)
300
LoI
CPP Studies:
Kosten et al. (1994)
200
Yates et al. (2011)
100
0
-100
-200
0.1
0.5
1.5
Amphetamine Dose (mg/kg)
Does Steep Discounting Precede
& Predict Drug Taking & Abuse?

Price elasticity of demand
Koffarnus & Woods (2013)
High-Impulsive (steep delay
discounting curves) rats inject more
cocaine than Low-Impulsive rats when
the price of cocaine increases.
Does Steep Discounting Precede
& Predict Drug Taking & Abuse?

Price elasticity of demand
Koffarnus & Woods (2013)
The effect is selective to drug-taking.
Summary
Human longitudinal data suggests steep
delay discounting precedes and predicts
drug taking.
 In rats, steep delay discounting
predicts…

 acquisition of cocaine taking but not other
drugs.
 willing to pay a higher price for cocaine (and
nicotine, Diergaarde et al., 2008).
Gregory J. Madden
Utah State University
Steep delay discounting is
correlated with obesity in women.
Davis et al. (2009)
See also Bruce et al. (2011), Epstein et al. (2013), Garza et al. (2012),
Jarmolowicz et al. (2014); Rasmussen et al. (2010), Reimers et al. (2009),
Stojek et al. (2014) & Weller et al. (2008)
Discounting rate is correlated
with health decision-making

Chabris et al. (2008)
 Steepness of delay discounting
function was a better predictor
of health-behavior and healthstatus (e.g., BMI) than were a
host of variables (e.g., age,
gender, depression, education,
and cognitive ability).
 See also Garza et al. (2012)
Irrational Decisions
Hyperbolic Discounting
Delay Discounting



What is it?
Why discount the future?
Why is it important?
 Part 1: Shape of the discounting function
predicts irrational choice.
 Part 2: Individuals differ in the degree to
which they devalue future consequences
and this correlates with addiction & health
decision-making.
Can we address this crisis with
Behavioral Economics?
Is it possible to produce a longlasting change in delay
discounting (an apparent
biobehavioral trait; Odum,
2011)?
Among children, how do we
influence health-related choice
knowing that steep discounting
is the baseline?
Goal: Long-lasting, Replicable Changes
to Delay Discounting

Shaping delay-tolerance
 Human
○ Short-term demonstration studies(Dixon & Holcomb,
2000; Schweitzer & Sulzer-Azaroff, 1988).
 Pigeons
○ Long-term effects (Mazur & Logue, 1978; Logue &
Mazur 1981)
○ Inconsistent results between studies (Logue et al.,
1984)
 Rats
○ Unable to replicate
Goal: Long-lasting, Replicable Changes
to Delay Discounting

Is it possible to produce a long-lasting change to a
biobehavioral trait (Odum, 2011)?
Reward bundling (Ainslie & Monterosso, 2003; Stein
et al., 2013)
100
Bundle-size 9
Bundle-size 3
Bundle-size 1
80
% LL Choice

60
40
20
0
0
5
7.5
10
Delay (s)
12.5
17.5
Goal: Long-lasting, Replicable Changes
to Delay Discounting
4
POST MAD: WMT & SHAM
3
25
20
2
r = .60; p < .01
0
Bickel et al. (2011)
2
Ln(MAD)
4
Renda et al. (2014)
MAD (s)

Is it possible to produce a long-lasting change to a
biobehavioral trait (Odum, 2011)?
Working-memory training
Ln(Retention Interval)

15
10
5
0
WMT
SHAM
Renda et al. (in prep)
Goal: Long-lasting, Replicable Changes
to Delay Discounting


Is it possible to produce a long-lasting change to a
biobehavioral trait (Odum, 2011)?
Delay-exposure (Stein et al., 2013)
Stein et al. (2013)
No-Delay Group (N=14)
Delay Group (N=14)
17.5 s delay
Stein et al. (2013)
Test of Impulsive Choice
Trial Block 1
Stein et al. (2013)
Test of Impulsive Choice
Trial Block 2
Effect held at 2 month follow-up
Replicated with new rats and at
longer delay.
Cocaine self-administration
Percent Larger Choice
15 s delay
100
50
17.5 s Delay
No Delay
0
0
15
Delay (s)
Goal: Long-lasting, Replicable Changes
to Delay Discounting

Brief Acceptance & Values Intervention
 Morrison et al. (2014)
○ Meet distressing
events with
acceptance
○ Progress in a
valued direction
with these events.
○ Waitlist control.
Can we address this crisis with
Behavioral Economics?
Is it possible to produce a longlasting change in delay
discounting (an apparent
biobehavioral trait; Odum,
2011)?
Among children, how do we
influence health-related choice
knowing that steep discounting
is the baseline?
I avoid
Cheetos
because of
health
concerns.
I love Cheetos!
I would eat
them all the
time if my
parents would
let me!
Repeated Tasting
Incentivize
Consumption
Pauline Horne
Fergus Lowe
Default Provision &
Increasing Labor Costs of Junk
Food
Wengreen et al. (2013)
Fruit
0.5
Vegetable
Cups Consumed
41% increase
0.4
36% increase
0.3
0.2
0.1
0.0
Bsln 1
Phase 2
Bsln 1
Incentivize
Consumption
Default Provision
Pauline Horne
Fergus Lowe
Phase 2
Difficulties Encountered
School had no budget for tangible incentives
and needed external labor to implement the
program.
 >40% of teachers regarded the intervention
as an “unfunded mandate” and did not
always implement the program as designed.


Cheating
Gamification

Using video game design
principles to influence
socially significant human
behavior.
Kevin Werbach’s MOOC (Wharton
School) on Gamification
Gamification
Using video game design
principles to influence
socially significant human
behavior.
 Playing games requires less labor
 Teachers like it (as do kids and
principals)
 There are no tangible rewards and
everyone is on the same team, so
cheating is minimized.

Gamification
Good vs. Evil Narrative
 Clear Object of the Game

The Good Guys
AKA: Role Models
The Bad Guys
Vegetation Annihilation
Team
Gamification
Good vs. Evil Narrative
 Clear Object of the Game
 Player Control

Because you met your fruit
and vegetable consumption
goal yesterday, here’s what
happened to the heroes…
Gamification
Good vs. Evil Narrative
 Clear Object of the Game
 Player Control
 Goldilocks-Zone Goal Setting

Most
Least
Did not meet your goal
Met your goal
C o n s u m p t io n ( P o r t io n s )
Jones et al. (2014a)
PLoS ONE
1 .0
B a s e lin e
G a m ific a tio n
B a s e lin e
C o m p e t it io n
p < .01
0 .8
p < .05
0 .6
0 .4
Fruit
0 .2
Veg
Fruit
Veg
+67%
+43%
0 .0
YEAR ONE
13-day intervention
YEAR TW O
28-day intervention
Fruit Consumption (gm)
100
Edith Bowen School
80
Jones et al. (2014b)
Preventive Medicine
60
+39%
40
20
0
Vegetable Consumption (gm)
Baseline
Target
NonTarget
100
80
60
40
+33%
20
0
Baseline
Target
NonTarget
Can we address this crisis with
Behavioral Economics?
Is it possible to produce a longlasting change in delay
discounting (an apparent
biobehavioral trait; Odum,
2011)?
Among children, how do we
influence health-related choice
knowing that steep discounting
is the baseline?
Gregory J. Madden
Utah State University