The date/delay effect in intertemporal choice: A combined fMRI and eye-tracking study

Abstract

Temporal discounting, the tendency to devalue future rewards as a function of delay until receipt, is influenced by time framing. Specifically, discount rates are shallower when the time at which the reward is received is presented as a date (date condition; e.g., June 8, 2023) rather than in delay units (delay condition; e.g., 30 days), which is commonly referred to as the date/delay effect. However, the cognitive and neural mechanisms of this effect are not well understood. Here, we examined the date/delay effect by analysing combined fMRI and eye-tracking data of N = 31 participants completing a temporal discounting task in both a delay and a date condition. The results confirmed the date/delay effect and revealed that the date condition led to higher fixation durations on time attributes and to higher activity in precuneus/PCC and angular gyrus, that is, areas previously associated with episodic thinking. Additionally, participants made more comparative eye movements in the date compared to the delay condition. A lower date/delay effect was associated with higher prefrontal activity in the date > delay contrast, suggesting that higher control or arithmetic operations may reduce the date/delay effect. Our findings are in line with hypotheses positing that the date condition is associated with differential time estimation and the use of more comparative as opposed to integrative choice strategies. Specifically, higher activity in memory-related brain areas suggests that the date condition leads to higher perceived proximity of delayed rewards, while higher frontal activity (middle/superior frontal gyrus, posterior medial frontal cortex, cingulate) in participants with a lower date/delay effect suggests that the effect is particularly pronounced in participants avoiding complex arithmetic operations in the date condition.

Keywords: date/delay effect; episodic thinking; eye-tracking; fMRI; precuneus; temporal discounting.

FIGURE 1

Illustration of the Intertemporal Choice Task. Each trial started with a fixation period. The duration was fixed at 6000 ms for the first trial and corresponded to the trial end fixation for the other trials. Afterwards, the decision screen (above: example of a delay trial; below: example of a date trial) was presented for 6000 ms during which a choice had to be made. Finally, a fixation cross was presented again. The duration of the latter stage was the sum of the time that would have been left during the decision screen (i.e., 6000 ms − reaction time [RT]) plus a jitter of 4000 ms to 8000 ms (drawn uniformly). All trials were presented in German (translation: Tage = days, Dezember = December). The font size was adjusted for better readability.

FIGURE 2

Number of Saccades Between Areas of Interest (AOIs) and Duration of Fixations Within AOIs. (a) represents all horizontal (Hor) and vertical (Ver) saccades between AOIs, and (b) represents saccades separated by attribute‐wise (Rew = reward), horizontal saccades and option‐wise (SIR, smaller‐immediate reward; LLR, larger‐later reward), vertical saccades. (c) represents the fixation duration on reward and time AOIs, summarised over options, and (d) represents fixation durations on reward and time AOIs, separated by options. Error bars represent standard errors of the mean. Asterisks are only depicted for post hoc t‐tests within the same category (i.e., bars next to each other). *p _adj < .05, ***p _adj < .001 (Bonferroni‐Holm corrected for multiple comparisons).

FIGURE 3

Blood‐oxygen level‐dependent (BOLD) signal during intertemporal choices. Depicted are significant areas from one‐sample t contrasts (a) delay > baseline and (b) date > baseline with an FWE‐corrected voxel‐related threshold (p < .05) and a minimum cluster size of 10 voxels. Intensities of colours reflect the heights of t‐values, as shown in the colour bar. Number labels refer to Montreal Neurological Institute (MNI) z‐coordinates.

FIGURE 4

Blood‐oxygen level‐dependent (BOLD) signal in Date > Delay. Depicted are significant areas from a one‐sample t contrast date > delay with a voxel‐related threshold of p < .001 and an FWE‐corrected cluster threshold. Intensity of colours reflects the heights of t‐values, as shown in the colour bar. For the delay > date contrast, no significant clusters emerged. Number labels refer to Montreal Neurological Institute (MNI) z‐coordinates.

FIGURE 5

Associations Between the Behavioural Date/Delay Effect and the Neural Date/Delay Effect. The behavioural date/delay effect (DDE), that is, ln(k)_Date–ln(k)_Delay, was used as covariate for the date > delay contrast (neural DDE). (a) represents significant areas at a voxel‐related threshold of p < .001 and an FWE‐corrected cluster threshold. Intensity of colours reflects the heights of t‐values, as shown in the colour bar. Upper labels refer to Montreal Neurological Institute (MNI) z‐coordinates. As the behavioural DDE was initially negatively coded (date minus delay), that is, lower values represent a stronger behavioural DDE, activity in these areas differed more strongly in people showing a smaller (or opposite) behavioural DDE. (b)–(f) illustrate the direction of the DDE‐dependent signal changes in significant clusters. For better interpretation, the x‐axis was recoded so that higher values represent a higher behavioural DDE (i.e., ln(k)_Delay–ln(k)_Date). ACC, anterior cingulate cortex; L, left; MCC, medial cingulate cortex; MFG, middle frontal gyrus; OCC, occipital regions; pMFC, posterior medial frontal cortex; R, right; SFG, superior frontal gyrus.