Discounting of Delayed Rewards: Effects of Response Format

1
Discounting of Delayed RewardsEffects of
Response Format Outcome Frame

• Laura Haynes
• Mike Aitken
• Trevor Robbins
• Behavioural Clinical Neurosciences Institute
• Dept. Experimental Psychology
• University of Cambridge

2
Temporal Discounting of Monetary Rewards
Current subjective value of delayed monetary
outcomes

• Discount fn exponential/hyperbolic?
• Hyperbolic devaluation across delay
• Value 1 / (1 kDelay)
• Discount rate characterised by k
• Larger k steeper discounting

3
Contrary to rational choice theories

• Descriptive Variance
• eg. framing effects
• eg. people require big bonuses to delay
  collection, but are only willing to sacrifice
  small sums to expedite it.
• Procedural Variance
• eg. discount rates choice gtgt matching

4
80 now OR 100 in 1 week?

• AIM1 Under matching conditions, do discount
  rates depend on whether response format is in
  days or dollars?
• How long would you wait to To delay collection
  1 week, how
• get an extra 20? much more would you want?
• AIM2 How does the effect of response format
  combine with framing effects to influence
  discount rates?

5
The Present Study

• 2x2x3 within subjects design (N32)
• Outcome frame (delay, expedite)
• X
• Response Format (days, )
• X
• Reward Magnitude (S, M, L)

6
Example Question

• Outcome Frame Delay
• Reponse Format
• You can collect a 80 cash prize now. If you
  were to delay collection by 1 week, what is the
  minimum cash bonus you would accept?

7
Results 1

• Magnitude effect
• Discount rates under delay frame were higher than
  under the expedite frame
• Response format
• Discount rates were higher when responses were
  in days than dollars

8
Results 2

• No interaction between factors of interest

9
Results 3 Conclusions

• Individual analyses showed that impulsivity on a
  choice task predicted the size of both the
  effects of frame and format.
• Outcome frame, response format and reward
  magnitude all exert strong but independent
  influences on discount rates.