Dissertation
What are the neural mechanisms of 'higher-order' timing?: complex behavior from low-level circuits
University of Iowa
Doctor of Philosophy (PhD), University of Iowa
Autumn 2021
DOI: 10.17077/etd.006279
Abstract
To act adaptively, we must predict future events and adjust our behavior accordingly. Predicting when upcoming events will occur is core to these expectations. We often generate temporal expectations based on cues that are available in the immediate environment. Experimenters often study this process by presenting subjects with a cue that instructs them to make a motor response after a certain time-interval has passed, such as training rats to press a lever 4 seconds after a tone is presented. These tasks have been used extensively to study the neural mechanisms of time perception, and consequently, neural models of timing often focus on ex-plaining these data.
However, in real-world environments, temporal cues likely provide more complex information than these tasks capture. Recent data suggests that the timing system has evolved to handle this complexity. For example, in a real-world setting, observers might encounter multiple cues that give conflicting information regarding when an event will occur, requiring that the timing system be able to integrate the information across cues. Consistent with this, when rats are presented with multiple cues that instruct them to respond after different intervals, they often time an average of their individual durations; an effect referred to as 'temporal averaging.' Furthermore, environmental conditions will often cause the intervals associated with different cues to covary. Consequently, when an individual learns that one cue's interval has changed, they would often benefit from generalizing this learning to other cues. Accordingly, when rats learn that one cue’s interval has changed, they act as if they expect other cue’s intervals have changed in the same direction—referred to as 'cross-cue transfer.'
Neural models have difficulty explaining these 'higher order' timing effects. However, we show that they readily emerge through biophysically realistic modifications to a ramp-to-threshold model proposed by Simen et al. (2011). In this model, temporal cues cause 'ramping units' to begin increasing their firing rates linearly as time passes. When they reach a 'threshold' firing rate, the observer will execute a response. To time responses appropriately, ramping units adjust their slopes to reach the threshold at a given cue’s interval (e.g., higher slopes for shorter intervals). In the original theory, temporal stimuli activate 'cue-units' that set the slope of the ramp by activating different numbers of 'tonic neurons.' Tonic units sent consistent excitation to the ramping neuron, and the more tonic units that are active the higher the ramp-slope and vice versa.
We show that temporal averaging emerges in the model when different cue-units compete over the number of tonic neurons that activate during a trial, via balanced excitation and inhibition. Furthermore, transfer emerges if learning-related plasticity occurs at the level of the tonic pool. Specifcally, if different cues share common tonic-units, adjustments to the ramp-slope from changing one cue’s interval will 'carry over' to cues that were never explicitly altered.In Chapter 2, we vet the model against existing behavioral literature, showing that the model accounts for the data well and finding several falsifiable predictions. In Chapter 3, we evaluate whether the mediodorsal nucleus of the thalamus might serve the role of delivering cue-related information to ramping units in the prefrontal cortex—a critical point to clarify to formally test our proposals. Consistent with this, inactivating either area in rats causes a profound disorganization of timing behavior, as if subjects lose access to the core timing circuit. Moreover, blocking communication between these areas yields the same effect. In Chapter 4, we test the specific hypothesis that, when duration-changes occur in different contexts, the hippocampus substitutes for the effects of plasticity at the tonic pool, allowing cross-cue transfer to be modulated by context. Consistent with this, rats with hippocampal lesions become less sensitive to contextual factors during transfer experiments. However, we found that the degree of hippocampal involvement differs depending on whether a cue is explicitly rewarded in different con-texts, suggesting that the model is currently incomplete.
Despite some limitations, the adapted model is currently the only neural theory of timing to our knowledge that collectively accounts for these effects and baseline components performance. Our proposals give simple, mechanistic explanations for how seemingly 'complex' timing behavior emerges from low-level circuits. Furthermore, the causal data provided here provides further avenues for falsifying the model.
Details
- Title: Subtitle
- What are the neural mechanisms of 'higher-order' timing?: complex behavior from low-level circuits
- Creators
- Benjamin John De Corte
- Contributors
- John H Freeman (Advisor)Krystal L Parker (Advisor)Edward A Wasserman (Committee Member)Eliot R Hazeltine (Committee Member)Daniel T Tranel (Committee Member)
- Resource Type
- Dissertation
- Degree Awarded
- Doctor of Philosophy (PhD), University of Iowa
- Degree in
- Neuroscience
- Date degree season
- Autumn 2021
- Publisher
- University of Iowa
- DOI
- 10.17077/etd.006279
- Number of pages
- xvii, 241 pages
- Copyright
- Copyright 2021 Benjamin John De Corte
- Language
- English
- Description illustrations
- color illustrations
- Description bibliographic
- Includes bibliographical references (pages 205-241).
- Public Abstract (ETD)
Time is a fundamental dimension of experience. Imagine if humans lacked any understanding of time. Could we communicate with one another if the words in our sentences were always spoken in a random order? Could early humans have found food if hunters did not understand that they should throw their spears before approaching their prey? Could Isaac Newton have begun formalizing our understanding of gravity if he could not grasp that the apple fell from the tree toward the earth rather than from the earth toward the tree? Finally, science as an institution seeks to discover causal relationships among phenomena in the natural world. Could it have developed if we could not comprehend that causes precede their effects? From these examples, one could argue that a basic representation of time is not only necessary for the survival of any species but also for advanced civilization. All events are embedded within time, and over the course of our evolution, our nervous systems have had to adapt to this fact. Consistent with this, we now know that a variety of species, ranging from ants to humans, represent time at one level or another. However, we still have a limited understanding of how the brain encodes this dimension. This is not for lack of trying, as we have many neural theories of timing. However, they often only capture basic forms of timing, failing to generalize to more ‘complex’ forms of time-perception. In this dissertation, we outline and test an adapted theory of timing that appears to resolve many of these problems.
- Academic Unit
- Interdisciplinary Graduate Program in Neuroscience
- Record Identifier
- 9984210749602771
Metrics
38 File views/ downloads
51 Record Views