Time perception and patience: individual differences in interval timing precision predict choice impulsivity in European starlings, Sturnus vulgaris

Abstract

Impulsivity, in the sense of the extent rewards are devalued as the time until their realization increases, is linked to various negative outcomes in humans, yet understanding of the cognitive mechanisms underlying it is limited. Variation in the imprecision of interval timing is a possible contributor to variation in impulsivity. We use a numerical model to generate predictions concerning the effect of timing imprecision on impulsivity. We distinguish between fixed imprecision (the imprecision that applies even when timing the very shortest time intervals) and proportional imprecision (the rate at which imprecision increases as the interval becomes longer). The model predicts that impulsivity should increase with increasing fixed imprecision, but decrease with increasing proportional imprecision. We present data from a cohort of European starlings (Sturnus vulgaris, n = 28) in which impulsivity had previously been measured through an intertemporal choice paradigm. We tested interval timing imprecision in the same individuals using a tri-peak temporal reproduction procedure. We found repeatable individual differences in both fixed and proportional imprecision. As predicted, birds with greater proportional imprecision in interval timing made fewer impulsive choices, whilst those with greater fixed imprecision tended to make more. Contradictory observations in the literature regarding the direction of association between timing imprecision and impulsivity might be clarified by distinguishing between fixed and proportional components of imprecision.

Keywords: Delay discounting; Impulsivity; Interval timing; Scalar expectancy theory; Starlings; Sturnus vulgaris.

Fig. 1

Two components of individual differences in timing imprecision. The imprecision in interval timing increases with the interval to be timed. The solid and dashed lines show this relationship for two hypothetical individuals. Here, one individual (dashed line) is more imprecise than another (solid line), in terms of a larger intercept of the line describing the magnitude of imprecision for a given interval duration (α, fixed imprecision), and also in terms of this line having a steeper slope (β, proportional imprecision)

Fig. 2

a Predicted effect from the numerical model of changes in the fixed component of timing imprecision (α) and the proportional component of timing imprecision (β) on the relative valuation of a smaller sooner reward (SS) after 3 s and a larger later reward after 8 s. Valuations are relative to those of an individual with perfectly precise timing. A higher relative valuation would lead to greater impulsivity. The discount rate is set at k = 0.54, an empirically derived estimate for starlings. b, c Illustrative explanations of the effects predicted by the model. b Additional imprecision that has the same magnitude at all delays (i.e. in α) has a stronger over-valuing effect on the smaller sooner reward than the larger later reward, because the discount function is steeper at this point. c Additional imprecision whose magnitude is proportional to the delay (i.e. in β) has a stronger over-valuing effect on the larger later reward than the smaller sooner reward, because the additional imprecision is greater at the longer delay

Fig. 3

a Overall tri-peak temporal reproduction procedure performance averaged across all birds. b, c Plots from two individual birds with differing interval timing performance. Fitted polynomial shown (solid line). The individual (BGPP) in panel b shows greater imprecision (Spread) as compared to the individual (BGYY) in panel c. Data show the mean number of responses on each key in 0.5 s time bins during PROBE trials during the final 4 days of the task. Vertical dashed reference lines show the FI durations on which birds were trained. Vertical solid grey lines indicate the geometric means of the first and second FI durations, and the second and third FI durations. Data for the remaining 25 birds are shown in Supporting Information, figure S2

Fig. 4

Spread of responses against fixed interval duration, by bird. Data are from the final 4 days of the tri-peak temporal reproduction task. Individual starlings are identified by a 4-letter code. Solid lines represent linear fits for each bird. Dotted lines represent expectation under the assumption of scalar variance, with the ratio of Spread to FI delay given by the grand mean across birds and delays

Fig. 5

a Relationship between fixed imprecision in timing and the proportion of choices for the smaller sooner reward (SS) in the impulsivity task. b Relationship between proportional imprecision in timing and the proportion of choices for the smaller sooner reward (SS) in the impulsivity task. Lines represent linear fits, and shaded areas 95% confidence intervals

Fig. 6

Effects of developmental treatments on timing imprecision. a Fixed imprecision. b Proportional imprecision. Developmental treatments are Effort: H is Hard, E is Easy; Amount: L is Lean, P is Plenty. Shown are means ± one standard error