Limits...
Cumulative Weighing of Time in Intertemporal Tradeoffs

View Article: PubMed Central - PubMed

ABSTRACT

We examine preferences for sequences of delayed monetary gains. In the experimental literature, two prominent models have been advanced as psychological descriptions of preferences for sequences. In one model, the instantaneous utilities of the outcomes in a sequence are discounted as a function of their delays, and assembled into a discounted utility of the sequence. In the other model, the accumulated utility of the outcomes in a sequence is considered along with utility or disutility from improvement in outcome utilities and utility or disutility from the spreading of outcome utilities. Drawing on three threads of evidence concerning preferences for sequences of monetary gains, we propose that the accumulated utility of the outcomes in a sequence is traded off against the duration of utility accumulation. In our first experiment, aggregate choice behavior provides qualitative support for the tradeoff model. In three subsequent experiments, one of which incentivized, disaggregate choice behavior provides quantitative support for the tradeoff model in Bayesian model contests. One thread of evidence motivating the tradeoff model is that, when, in the choice between two single dated outcomes, it is conveyed that receiving less sooner means receiving nothing later, preference for receiving more later increases, but when it is conveyed that receiving more later means receiving nothing sooner, preference is left unchanged. Our results show that this asymmetric hidden-zero effect is indeed driven by those supporting the tradeoff model. The tradeoff model also accommodates all remaining evidence on preferences for sequences of monetary gains.

No MeSH data available.


Observed proportions of participants choosing the high NPV option among those lending anecdotal support to the respective models (horizontal), plotted against the average predicted probabilities of choosing the high NPV option generated by the models (vertical), in Experiment 2. TM = Tradeoff Model; DIUM = Discounted Instantaneous Utility Model; SMNZ = No-Zero Sequences Model; SMZ = Zero Sequences Model.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4998108&req=5

fig3: Observed proportions of participants choosing the high NPV option among those lending anecdotal support to the respective models (horizontal), plotted against the average predicted probabilities of choosing the high NPV option generated by the models (vertical), in Experiment 2. TM = Tradeoff Model; DIUM = Discounted Instantaneous Utility Model; SMNZ = No-Zero Sequences Model; SMZ = Zero Sequences Model.

Mentions: Bayes Factors and posterior probabilities inform us about the relative performance of the models, not about how well each model describes the choices of those best described by it. Figure 3 depicts the observed proportions of participants choosing the high NPV option among those lending anecdotal support to the respective models in the main contest (horizontal), plotted against the average predicted probabilities of choosing the high NPV option generated by the models (vertical). Not surprisingly, each model describes the choices of its supporters best. Interesting is how TM and DIUM describe the supporters of one another. Most choice pairs, except for #1 to #4, were selected for generating an ordinal contrast between preference entailed by duration (TM, with linear utility) and preference entailed by Net Present Value (DIUM, with linear utility). Table 3 shows that supporters of TM indeed mostly prefer the low NPV option, whereas supporters of DIUM mostly prefer the high NPV option, and Figure 3 shows that TM tends to underpredict high NPV choices among supporters of DIUM, whereas DIUM tends to overpredict high NPV choices among supporters of TM (as the sequences model do as well).


Cumulative Weighing of Time in Intertemporal Tradeoffs
Observed proportions of participants choosing the high NPV option among those lending anecdotal support to the respective models (horizontal), plotted against the average predicted probabilities of choosing the high NPV option generated by the models (vertical), in Experiment 2. TM = Tradeoff Model; DIUM = Discounted Instantaneous Utility Model; SMNZ = No-Zero Sequences Model; SMZ = Zero Sequences Model.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4998108&req=5

fig3: Observed proportions of participants choosing the high NPV option among those lending anecdotal support to the respective models (horizontal), plotted against the average predicted probabilities of choosing the high NPV option generated by the models (vertical), in Experiment 2. TM = Tradeoff Model; DIUM = Discounted Instantaneous Utility Model; SMNZ = No-Zero Sequences Model; SMZ = Zero Sequences Model.
Mentions: Bayes Factors and posterior probabilities inform us about the relative performance of the models, not about how well each model describes the choices of those best described by it. Figure 3 depicts the observed proportions of participants choosing the high NPV option among those lending anecdotal support to the respective models in the main contest (horizontal), plotted against the average predicted probabilities of choosing the high NPV option generated by the models (vertical). Not surprisingly, each model describes the choices of its supporters best. Interesting is how TM and DIUM describe the supporters of one another. Most choice pairs, except for #1 to #4, were selected for generating an ordinal contrast between preference entailed by duration (TM, with linear utility) and preference entailed by Net Present Value (DIUM, with linear utility). Table 3 shows that supporters of TM indeed mostly prefer the low NPV option, whereas supporters of DIUM mostly prefer the high NPV option, and Figure 3 shows that TM tends to underpredict high NPV choices among supporters of DIUM, whereas DIUM tends to overpredict high NPV choices among supporters of TM (as the sequences model do as well).

View Article: PubMed Central - PubMed

ABSTRACT

We examine preferences for sequences of delayed monetary gains. In the experimental literature, two prominent models have been advanced as psychological descriptions of preferences for sequences. In one model, the instantaneous utilities of the outcomes in a sequence are discounted as a function of their delays, and assembled into a discounted utility of the sequence. In the other model, the accumulated utility of the outcomes in a sequence is considered along with utility or disutility from improvement in outcome utilities and utility or disutility from the spreading of outcome utilities. Drawing on three threads of evidence concerning preferences for sequences of monetary gains, we propose that the accumulated utility of the outcomes in a sequence is traded off against the duration of utility accumulation. In our first experiment, aggregate choice behavior provides qualitative support for the tradeoff model. In three subsequent experiments, one of which incentivized, disaggregate choice behavior provides quantitative support for the tradeoff model in Bayesian model contests. One thread of evidence motivating the tradeoff model is that, when, in the choice between two single dated outcomes, it is conveyed that receiving less sooner means receiving nothing later, preference for receiving more later increases, but when it is conveyed that receiving more later means receiving nothing sooner, preference is left unchanged. Our results show that this asymmetric hidden-zero effect is indeed driven by those supporting the tradeoff model. The tradeoff model also accommodates all remaining evidence on preferences for sequences of monetary gains.

No MeSH data available.