Scientific research in Portugal on the perception of time
A team of researchers from the Champalimaud Research Learning Lab, made up of Tiago Monteiro, Filipe S. Rodrigues, Margarida Pexirra, Bruno F. Cruz, Ana I. Gonçalves, Pavel E. Rueda-Orozco and Joseph J. Paton, recently published a paper in the journal Nature Neuroscience, in which they describe how they artificially slowed down or sped up patterns of neuronal activity in rats, altering their judgement of the passage of time and presenting the most persuasive causal evidence available to date on how the brain works in this respect. The paper is entitled: Using temperature to analyze the neural basis of a time-based decision.
Joseph Paton and his colleagues trained a group of rats to distinguish between different time intervals. During their research, they observed that activity in the corpus striatum (striatum), a deep region of the brain, showed predictable patterns that varied at different rates.
To test this tool in rats, the researchers developed a custom thermoelectric device that allows them to locally warm or cool the striatum while recording neuronal activity. In these experiments the rats were anaesthetised, so the researchers used optogenetics - a technique that uses light to stimulate specific cells - to create waves of activity in the corpus striatum.
The animals were trained to indicate whether the interval between two tones was longer or shorter than 1.5 seconds.
When the striatum was cooled, the animals were more likely to interpret a given interval as short. On the other hand, when the striatum was warmed, they were more likely to indicate that the interval was long.
Warming the striatum accelerated the population dynamics of that brain region, much like speeding up the movement of a clock's hands, which led the rats to judge a given time interval as longer than it really was.
The team's findings indicate that the striatum plays a fundamental role in determining what to do and when to do it. However, the control of movement is related to other brain structures.
By providing new insight into the causal relationship between neuronal activity and timing function, the team's findings represent a significant step forward in the development of new therapeutic targets for debilitating diseases such as Parkinson's disease or Huntington's disease, which present symptoms related to time and to the striatum. This new information could potentially pave the way for more effective and targeted interventions to treat these complex neurological diseases, thereby improving the quality of life of the patients affected by them.
THE POPULATION CLOCK
This is a theory referred to in this paper, from the field of neurobiology, which suggests that the human brain does not have a single centralised clock for processing time, but rather multiple "clocks" distributed across different neuronal networks. According to this hypothesis, different groups of neurons in various regions of the brain are involved in processing time. These neuronal groups would form a "population clock", which relies on patterns of neuronal activity to calculate and represent the passage of time in a decentralised and flexible way.
This approach differs from the concept of a "centralised clock", such as the one used in computers, in which a single source controls timekeeping. In the brain, the perception of time seems to emerge from the complex interaction between several brain areas and their coordinated neuronal activities.
The "population clock" hypothesis remains the subject of study and research in neurobiology, and understanding it may contribute to a better understanding of how human beings perceive and process time.
THE PAPER
The full paper can be read at this link. Nevertheless, here is a translation of the introduction and of part of the conclusion, for readers most eager for knowledge.
ABSTRACT
"The basal ganglia are thought to contribute to decision-making and motor control. These functions depend critically on temporal information, which can be extracted from the evolving state of neuronal populations in their main input structure, the striatum. However, it is debated whether striatal activity underlies latent, dynamic decision processes or the kinematics of overt movement. Here, we measured the impact of temperature on striatal population activity and on the behaviour of rats, and the observed effects were compared with neuronal activity and behaviour collected in several versions of a temporal categorisation task. Cooling caused dilation, and warming contraction, of both neuronal activity and patterns of judgement in time, mimicking endogenous decision-related variability in striatal activity. However, temperature did not similarly affect movement kinematics. These data provide compelling evidence that the time course of evolving striatal activity dictates the speed of a latent process that is used to guide choices, but not continuous motor control. More broadly, they establish the timescale of population activity as a likely neuronal basis for variability in timing behaviour."
INTRODUCTION
"(...) Much of behaviour depends on time. Humans and other animals must extract the temporal structure of the environment in order to learn to anticipate events, to understand the relationships between actions and consequences, and to estimate time, implicitly or explicitly, in order to plan, sequence and properly coordinate actions. For tasks as varied as waiting at a traffic light or a hummingbird foraging for nectar, time is fundamental. Timing mechanisms appear to be distributed throughout the nervous system, reflecting the importance of temporal information for much of brain function. However, a requirement common to diverse functions is the need to create an index, ordering and spacing information along the temporal dimension, so that useful relationships can be extracted and outputs properly coordinated. On the scale of seconds and minutes over which much of behaviour unfolds, neuronal population dynamics represent a candidate means both for encoding and for generating temporal patterns. Artificial neural network models have explored the evolution of population activity as a basis for timing sensory events and movements, and correlations between behaviour and the time course of population activity have lent some support to the hypothesis that time-varying patterns of activity within a population perform temporal computations. However, correlation does not imply causation. The critical prediction of these "population clock" hypotheses is that interventions capable of slowing down or speeding up the temporal evolution of activity should cause a corresponding dilation or contraction of the temporal functions performed with that activity. One brain system in which temporal information appears to be critical is that of the basal ganglia (BG), an evolutionarily ancient set of brain structures thought to contribute to the selection of appropriate actions on the basis of experience. A dominant view holds that the BG embodies essential features of reinforcement learning (RL) algorithms. In mammals, inputs from a diverse set of territories of the cortex, the thalamus and limbic brain structures convey information about the state of the world that converges onto the dense dopaminergic innervation in the main input area of the BG, the striatum. Input from dopaminergic neurons is thought to teach striatal circuits the value of taking certain actions in a given state, information that can be passed on to downstream brainstem and thalamocortical motor circuits to influence the selection or specify the features of actions. To perform such a function, the BG would need access to information about ordering and spacing along the temporal dimension, implicitly or explicitly, both to extract meaningful relationships between environment, actions and outcomes that drive learning and to coordinate the production of actions in time. Interestingly, data from people with disorders of the central nervous system and functional magnetic resonance imaging (fMRI) have consistently identified the central nervous system as being involved in timing behaviour. Furthermore, lesions and pharmacological manipulations of the striatum can cause deficits in temporal estimation and reproduction. Finally, recordings of striatal populations have shown that temporal information can be decoded from neural activity and that this information is correlated with variability in timing behaviour. Specifically, the state of striatal population activity changes continuously along reproducible trajectories during behavioural tasks that require the estimation of time, advancing more quickly when animals report long judgements and more slowly when they report short judgements. To test whether variability in the speed of BG population dynamics is merely correlated with, or directly regulates, timing function, we sought to experimentally manipulate the dynamics as animals reported temporal judgements. Interestingly, despite being composed of elements with different temperature dependencies (for example, ion channel conductances and synaptic transmission), neural circuits in at least some systems can produce activity patterns that systematically slow down or speed up with decreasing or increasing temperature. For this reason, temperature manipulations offer a potential method for testing hypotheses about the relationship between the speed of neural dynamics and function. Indeed, temperature manipulations in the zebra finch were used to identify the HVC area (proper name) as a locus in the song production circuit that contributes to the temporal pattern of birdsong. Similar temperature manipulations identified a subregion of the human motor cortex that regulates the speed of speech and a region of the rodent medial frontal cortex that controls the timing of a delayed movement. However, temperature can have distinct effects on neural activity, depending on the cellular and circuit features of the area in question.
Here, we used a custom thermoelectric device (TED) to systematically vary the temperature of striatal tissue, both warming and cooling relative to a baseline condition. We found that temperature affected overall activity levels non-monotonically, with both warming and cooling relative to a control temperature producing lower overall firing rates. In contrast, temperature manipulations caused monotonic, graded changes in the timescale of neural activity, mimicking the decision-related variability in activity observed during multiple versions of a temporal judgement task. Temperature manipulations also caused bidirectional, graded changes in the animals' temporal judgements, mirroring the temperature-dependent modification of the timescale of neural activity, as well as the observed relationship between the timescale of activity and the animals' judgements. Strikingly, these results were not accompanied by similar effects of temperature on movement execution. Instead, although more modest than the effects of temperature on the animals' judgements, the pattern of average speed of the animals' movements was a non-monotonic function of temperature, similar to the observed non-monotonic effect of temperature on baseline firing rate. Together, these results imply that distinct aspects of behaviour may be controlled by the temporal evolution of population activity and by overall activity levels. These distinctions suggest that continuously evolving activity patterns in the striatum support the placement of discrete behavioural transitions in time, as required for action selection and decision-making, but do not provide signals related to moment-to-moment motor commands needed for continuous control. Rather than the BG providing high-dimensional continuous control signals, the overall level of activity appears to provide a low-dimensional gain signal applied to motor programmes executed elsewhere, which is consistent with the role of BG circuits in modulating the vigour of movement.
CONCLUSION
(...) Previous work has shown that the speed of neural population activity along reproducible trajectories in neural space can correlate with variability in the timing of actions and with time-dependent decisions. Indeed, there is growing evidence that this may represent a general principle for temporal processing in the brain. Here we provide additional causal evidence for this principle, showing that experimental temperature manipulations in the striatum can be used to slow down or speed up both patterns of neural activity under anaesthesia and the temporal evolution of latent decision variables that rats use to guide behaviour, specifically their categorical judgements of duration. These results seem to reflect a direct impact of temperature on decision-related (or action-selection-related) processes and not on processes related to moment-to-moment movement execution (or to continuous control), because the effects of temperature on timing judgements and on movement kinematics were qualitatively distinct. It has long been known, in fields as diverse as neuroscience, robotics and artificial intelligence, that movement control is probably facilitated by a hierarchy of control mechanisms. Even relatively simple nervous systems must switch between distinct motor pattern generators as well as continuously control their output. The observation that an intervention capable of modulating striatal population speed has a differential impact on the evolution of decision variables and on the variables needed for continuous movement control is consistent with the proposal that the BG acts as an intermediate-level controller, important for selecting between, switching on or modulating different actions, but provides evidence against it issuing the moment-to-moment commands needed for movement execution. One caveat to this interpretation is that any potential movements occurring during the interval period, when the animals' judgements were presumably forming, were below our detection threshold. We therefore analysed the kinematics of choice movements, which necessarily occurred after the temporal judgement was complete. It is thus possible that temperature produced similar effects on judgements and on the kinematics of undetected putative movements during the decision process, and that its effects on movement kinematics changed qualitatively when it came to executing choice movements a few hundred milliseconds later. We consider this possibility less likely for reasons of parsimony and because of existing data that seem inconsistent with the BG providing detailed moment-to-moment kinematic information for use in behavioural control. For example, lesions of the dorsal striatum have been shown to disrupt the sequencing of behavioural motifs without affecting their execution. Even data showing that striatal lesions can affect the detailed kinematics of learned stereotyped movements can be reconciled with a more specific role of the BG in discrete control, since learned movements appear to involve the concatenation of motor elements, or sequencing, which implies that discrete transitions must be appropriately placed within the overall time course of behaviour. The observations that kinematic information can be decoded from the striatum during tasks involving timing are also consistent with this view, because BG circuits would probably benefit from access to continuous information about the state of the commands being sent to effector systems to guide when to produce certain actions, even though they are not responsible for sending detailed kinematic control signals to downstream circuits. Our results are reminiscent of the observation that cooling an orofacial region of the motor cortex in singing mice slows certain aspects of song timing while leaving the duration of individual notes unchanged. Interestingly, tonic inhibition of the direct feed pathway of the https://doi.org/10.1038/s41593-023-01378-5. What is interesting is that tonic inhibition of the direct feed pathway of the BG at its point of initiation in the striatum can produce a slowing of movement, which suggests that the influence of the BG on vigour-related control parameters may not be exerted through dynamics in the higher-dimensional space of population firing, but rather through a low-dimensional, gain-like modulation of motor programmes that are implemented largely by circuits elsewhere. In line with this, we found that endogenous variability in the baseline firing rate of striatal neurons was related to reaction times and that the non-monotonic effect of striatal temperature on this feature resembled a pattern in the rats' behaviour, in which both warmer and colder temperatures produced slower and delayed choice movements. During behaviour, animals interact continuously with the environment, and this interaction drives substantial amounts of neural activity in many brain areas. Identifying features of neural activity underlying latent processes, such as aspects of cognition, as opposed to behaviour-driven neural activity, is a difficult problem that requires multiple approaches. One approach involves studying neural activity in behavioural tasks that vary along dimensions orthogonal to the process of interest. For example, signals related to a decision rather than to its report should be invariant to the movement conditions of the task. In the present study, we found that variability in the speed of striatal population responses correlated with the subjects' temporal judgements in two variants of the task that differed in the degree to which subjects were free to move during stimulus presentation. This consistency argues against a scenario in which movement is the main driver of the correlation between neural variability and temporal judgement. A priori, it is impossible to determine whether such correlations reflect a causal relationship between that neural activity and behaviour, an unobserved source of neural activity elsewhere that directly causes both striatal neural activity and behaviour, or potentially behaviour causing striatal neural activity through re-entrant sensory input. Previous analysis of high-speed video during the fixation-free version of the temporal judgement task we study here showed that decision-related information appeared in striatal activity several hundred milliseconds before it appeared in overt behaviour, arguing against decision signals being driven solely by the sensory consequences of behaviour. Nevertheless, the existence of such varied possibilities underlines the need for so-called causal interventions to help determine whether neural activity is a cause, a corollary or a consequence of behaviour. Here, to examine the impact of a temperature intervention on population dynamics, we studied the dynamics evoked by optogenetic activation of the VB thalamus under anaesthesia, rather than the endogenous patterns of activity observed during behaviour. This approach has the disadvantage of not allowing 1) the simultaneous assessment of the impact of temperature on neural activity and on behaviour, and 2) the neural activity being generated through circuit mechanisms that are distinct from those at play during behaviour. However, it also has advantages. Specifically, behavioural sources of neural variability are removed entirely, allowing a much more direct assessment of the effects of temperature on the features of neural population dynamics that were independently correlated with timing judgements in several sets of experiments. In terms of delineating the nature of the causal influences among the various potential factors at play, technically this independence strengthens our confidence that the causal chain runs from temperature, to population dynamics, to the animals' temporal judgements. What circuit mechanisms might give rise to the task-relevant striatal population activity studied here? Earlier views suggested that the BG largely inherited their activity patterns from their inputs, often hypothesised to originate in the cortex. However, if the time-varying striatal activity underlying temporal judgement were simply inherited from another brain area, one would expect manipulations of striatal temperature to produce a minimal shift rather than a significant rescaling of behaviour. As in the vocal control circuit of songbirds, our data thus appear inconsistent with a mechanism in which the relevant dynamics are simply inherited by the brain area targeted by temperature manipulations, in our case the DS. In songbirds, there is evidence that a combination of local circuit mechanisms in the pallial area HVC and a larger re-entrant circuit involving HVC and several other brain areas are involved in generating the temporally patterned activity underlying song timing. A similar scenario may underlie the mechanisms generating temporally patterned striatal activity involved in temporal judgements. First, although most network modelling efforts that use neural dynamics for computation rely, at least in part, on recurrent excitation, recent work suggests that it may be possible for a largely inhibitory, striatum-like circuit to produce complex spatiotemporal dynamics given sustained excitatory input. The striatum may also represent one stage of a wider re-entrant circuit involving several brain systems, in which the wider circuit contributes to generating dynamic patterns of activity that govern the evolution of decision variables. From this perspective, delays or advances induced by cooling or warming would accumulate with each cycle through the circuit, resulting in a temporal rescaling with temperature. Such a circuit could, in principle, involve the cortex, BG structures and the thalamus, or subcortical areas such as downstream BG structures, the superior colliculus and the thalamus. However, our data suggest that any re-entrant circuit mechanism involving the cortex did not involve the primary motor cortex, since temperature manipulations had negligible effects on choice behaviour, consistent with previous studies showing that manipulating the motor cortex does not affect expert behaviour in motor timing tasks. Nevertheless, orbitofrontal and medial frontal cortical areas have been shown to encode temporal information during motor timing and temporal judgement tasks, albeit less precisely than the striatum, and cooling of medial frontal cortical structures has been shown to delay movements, potentially indicating the involvement of frontal cortical structures. Furthermore, the activity of midbrain dopaminergic neurons is correlated with, and can directly cause, changes in temporal judgements, suggesting that dopaminergic neuromodulation may also adjust the time course of network activity through its action on striatal circuits. It has been proposed that timing processes are distributed in the brain and that networks of neurons implicitly possess a rich capacity to act as timing mechanisms through the time-varying patterns of activity they tend to produce during behaviour, sometimes referred to as a "population clock". Although the data presented here strongly support this hypothesis, in principle the kinds of computations performed by earlier, more algorithmic and information-processing accounts of timing may well be embodied in the kind of population activity we describe here. This possibility is reflected in a growing belief that the brain performs many of its computations through dynamics. Our perceptions, thoughts and actions are continuously interlinked and regulated in time, and understanding the neural basis of temporal processing has been advocated as a necessary prerequisite for general models of cognition. However, understanding how the brain properly orders and spaces information along the temporal dimension in order to encode signals that are functions of time has been an enduring challenge for neuroscience. In this paper, we present compelling evidence that the time course of activity in populations of striatal neurons directly influences the time course of a timing process used to guide decision-making. The data not only imply a causal link between the timescale of a population response and a temporal basis for computation in the brain, but also argue that, in the hierarchy of behavioural control, striatal dynamics may act at an interface between cognition and motor function to help guide whether and when to produce which actions, but not to provide the continuously evolving, moment-to-moment motor commands needed for movement execution. Understanding the precise circuit mechanisms responsible for establishing and modulating the timescale of neural activity in these circuits, and which specific computations this activity serves, represent important future directions for understanding how the brain operates flexibly on internally computed information to produce adaptive and intelligent behaviour.
Comments