The timing database: An open-access, live repository for interval timing studies

Time-scale invariant contingency yields one-shot reinforcement learning despite extremely long delays to reinforcement

Reinforcement learning inspires much theorizing in neuroscience, cognitive science, machine learning, and AI. A central question concerns the conditions that produce the perception of a contingency between an action and reinforcement-the assignment-of-credit problem. Contemporary models of associative and reinforcement learning do not leverage the temporal metrics (measured intervals). Our information-theoretic approach formalizes contingency by time-scale invariant temporal mutual information. It predicts that learning may proceed rapidly even with extremely long action-reinforcer delays. We show that rats can learn an action after a single reinforcement, even with a 16-min delay between the action and reinforcement (15-fold longer than any delay previously shown to support such learning). By leveraging metric temporal information, our solution obviates the need for windows of associability, exponentially decaying eligibility traces, microstimuli, or distributions over Bayesian belief states. Its three equations have no free parameters; they predict one-shot learning without iterative simulation.

Creating a Home for Timing Researchers: Then, Now, and the Future

Our ability to perceive event duration and order is critical in every aspect of our lives, from everyday tasks like coordinating our limbs to walk safely, to uniquely human activities like planning our children's future. Many theoretical accounts of timing have been proposed to explain the mechanisms underlying our ability to estimate time and unify events in time. Continuous progress is being met in further refining and extending current theories, with the aim not only to advance our understanding of timing and time perception, but also to make timing more accessible and applicable to daily life. For this to be possible, cross-disciplinary thinking is required, which is something one cannot easily attain in a scientific conference, rather it requires a community. Having a community with an interest and/or expertise in timing can allow for cross-fertilization of ideas. This chapter introduced the story of the Timing Research Forum or else TRF.

Core clock gene, Bmal1, is required for optimal second-level interval production

Perception and production of second-level temporal intervals are critical in several behavioral and cognitive processes, including adaptive anticipation, motor control, and social communication. These processes are impaired in several neurological and psychological disorders, such as Parkinson's disease and attention-deficit hyperactivity disorder. Although evidence indicates that second-level interval timing exhibit circadian patterns, it remains unclear whether the core clock machinery controls the circadian pattern of interval timing. To investigate the role of core clock molecules in interval timing capacity, we devised a behavioral assay called the interval timing task to examine prospective motor interval timing ability. In this task, the mouse produces two separate nose pokes in a pretrained second-level interval to obtain a sucrose solution as a reward. We discovered that interval perception in wild-type mice displayed a circadian pattern, with the best performance observed during the late active phase. To investigate whether the core molecular clock is involved in the circadian control of interval timing, we employed Bmal1 knockout mice (BKO) in the interval timing task. The interval production of BKO did not display any difference between early and late active phase, without reaching the optimal interval production level observed in wild-type. In summary, we report that the core clock gene Bmal1 is required for the optimal performance of prospective motor timing typically observed during the late part of the active period.

Dynamics of retrospective timing: A big data approach

Most interval timing research has focused on prospective timing tasks, in which participants are explicitly asked to pay attention to time as they are tested over multiple trials. Our current understanding of interval timing primarily relies on prospective timing. However, most real-life temporal judgments are made without knowing beforehand that the durations of events will need to be estimated (i.e., retrospective timing). The current study investigated the retrospective timing performance of ~24,500 participants with a wide range of intervals (5-90 min). Participants were asked to judge how long it took them to complete a set of questionnaires that were filled out at the participants' own pace. Participants overestimated and underestimated durations shorter and longer than 15 min, respectively. They were most accurate at estimating 15-min long events. The between-subject variability in duration estimates decreased exponentially as a function of time, reaching the lower asymptote after 30 min. Finally, a considerable proportion of participants exhibited whole number bias by rounding their duration estimates to the multiples of 5 min. Our results provide evidence for systematic biases in retrospective temporal judgments, and show that variability in retrospective timing is relatively higher for shorter durations (e.g., < 30 min). The primary findings gathered from our dataset were replicated based on the secondary analyses of another dataset (Blursday). The current study constitutes the most comprehensive study of retrospective timing regarding the range of durations and sample size tested.