Project 1 - Fundamental Insight
Habituation and Novelty Detection
Practical challenges of directly observing and intervening in underlying plasticity as it occurs in learning animals have been major barriers to progress. Overcoming these barriers requires study of robust and reliable forms of learning and memory that are equally important across species, and which produce pronounced plasticity in regions where cell types and circuits are well understood, and structure and function is conserved across mammals. An additional advantage is provided where plasticity is so pronounced that it can be observed within relevant circuitry, making a wide range of methods to assess synaptic function feasible. To this end, we study habituation, a form of learning that is essential for survival and well-being as it filters out irrelevant stimuli and reduces their burden on attention and energy, enabling organisms to focus on signals of reward or punishment, or novel features that require exploration.
Stimulus-selective Response Plasticity
We study eye-specific, orientation-selective visual habituation, which produces pronounced potentiation of responses in primary visual cortex (V1) through a process known as stimulus-selective response plasticity (SRP). After mice view a phase reversing, oriented sinusoidal grating stimulus, neural responses in layer 4 of V1 increase dramatically with each day of viewing the same orientation, just as behavioural responses diminish through long-term habituation. After several days viewing this orientation, the stimulus-specificity of SRP and habituation is revealed by introducing interleaved presentations of a novel stimulus orientation. This novel stimulus returns cortical and behavioural responses to baseline magnitudes, revealing striking orientation-selectivity. SRP manifests as a large increase in the magnitude of visual evoked potentials (VEP) or in the peak ring rate of action potentials recorded in thalamo-recipient layer 4. However, SRP reflects a temporal redistribution of action potentials as peak ring rate is increased to phase reversals of the familiar stimulus but is less sustained when compared to novel stimuli (see figure below).
Stimulus-selective Response Plasticity
We study eye-specific, orientation-selective visual habituation, which produces pronounced potentiation of responses in primary visual cortex (V1) through a process known as stimulus-selective response plasticity (SRP). After mice view a phase reversing, oriented sinusoidal grating stimulus, neural responses in layer 4 of V1 increase dramatically with each day of viewing the same orientation, just as behavioural responses diminish through long-term habituation. After several days viewing this orientation, the stimulus-specificity of SRP and habituation is revealed by introducing interleaved presentations of a novel stimulus orientation. This novel stimulus returns cortical and behavioural responses to baseline magnitudes, revealing striking orientation-selectivity. SRP manifests as a large increase in the magnitude of visual evoked potentials (VEP) or in the peak ring rate of action potentials recorded in thalamo-recipient layer 4. However, SRP reflects a temporal redistribution of action potentials as peak ring rate is increased to phase reversals of the familiar stimulus but is less sustained when compared to novel stimuli (see figure below).
[Picture]
Layer 4 response magnitude in V1 diminishes with stimulus repetition within session, but across days there is a dramatic increase leading to biased response to familiar over novel stimuli. Figure adapted from Cooke et al. 2015 and Hayden et al. 2023
NMDA receptor dependency
The NMDA class of glutamate receptor plays a key role in many forms of Hebbian synaptic plasticity and learning. It is an ionotropic receptor that conducts calcium ions, a canonical trigger for many signaling mechanisms required for Hebbian synaptic plasticity, and it has the dual properties of ligand- and voltage-dependency, which make it an ideal detector of coincidence between pre- and post-synaptic activity, which is a defining criterion for Hebbian potentiation. We have demonstrated that the NMDA receptor is necessary for long-term habituation and accompanying physiological effects across different timescales, including within-session adaptation and stimulus-selective response potentiation over days. Thus, it appears that there is an involvement of Hebbian synaptic plasticity in these processes. Through local genetic knockdown of the NMDA receptor limited to binocular primary visual cortex (V1), we have demonstrated that these forms of plasticity rely on NMDA receptors local to V1 (see figure below). Our more recent work has started to target the specific cell types within which NMDA receptors are required, providing a stronger indication of selective circuits involved in these forms of plasticity across different timescales.
The NMDA class of glutamate receptor plays a key role in many forms of Hebbian synaptic plasticity and learning. It is an ionotropic receptor that conducts calcium ions, a canonical trigger for many signaling mechanisms required for Hebbian synaptic plasticity, and it has the dual properties of ligand- and voltage-dependency, which make it an ideal detector of coincidence between pre- and post-synaptic activity, which is a defining criterion for Hebbian potentiation. We have demonstrated that the NMDA receptor is necessary for long-term habituation and accompanying physiological effects across different timescales, including within-session adaptation and stimulus-selective response potentiation over days. Thus, it appears that there is an involvement of Hebbian synaptic plasticity in these processes. Through local genetic knockdown of the NMDA receptor limited to binocular primary visual cortex (V1), we have demonstrated that these forms of plasticity rely on NMDA receptors local to V1 (see figure below). Our more recent work has started to target the specific cell types within which NMDA receptors are required, providing a stronger indication of selective circuits involved in these forms of plasticity across different timescales.
[Picture]
Local knockdown of NMDA receptors in V1 using viral vectors produces a significant impairment of SRP, revealing both that it is a local effect in V1 and that it relies upon the NMDA receptor. Adapted from Cooke et al. 2015
Different classes of GABAergic inhibitory neurons contribute to habituation over different timescales
Using a variety of approaches, we have shown that Parvalbumin-expressing (PV+) GABAergic inhibitory neurons play a critical role in SRP: Activation or suppression of activity in these neurons prevents differential response to familiarity and novelty, both cortically and behaviourally. Calcium imaging of layer 4 PV+ neurons reveals not only that they are highly engaged by novel stimuli, but also that their activity decreases dramatically over days with long-term familiarity, and eventually is inhibited below baseline activity levels. Critically, calcium imaging of Somatostatin-expressing (SST+) neurons, another subtype of inhibitory neuron, shows that they become progressively engaged as stimuli become familiar over days and are suppressed by novel stimuli, in striking opposition to PV+ neuronal activity (see figure below). Consistent with a current understanding that SST+ neurons inhibit PV+ neurons in layer 4, we propose that long-term habituation is mediated by the influence of memory on SST+ neurons, which both inhibit excitatory neuronal targets through dendritic inhibition but also disinhibit those same targets by inhibiting PV+ neurons. Indeed, optogenetic stimulation of SST+ inhibitory neurons in V1 inhibits PV+ neurons to produce net disinhibition of cortical neurons. Thus, we are gaining a deep understanding of unexpected but critical circuit level interactions that support a foundational form of learning and memory.
Using a variety of approaches, we have shown that Parvalbumin-expressing (PV+) GABAergic inhibitory neurons play a critical role in SRP: Activation or suppression of activity in these neurons prevents differential response to familiarity and novelty, both cortically and behaviourally. Calcium imaging of layer 4 PV+ neurons reveals not only that they are highly engaged by novel stimuli, but also that their activity decreases dramatically over days with long-term familiarity, and eventually is inhibited below baseline activity levels. Critically, calcium imaging of Somatostatin-expressing (SST+) neurons, another subtype of inhibitory neuron, shows that they become progressively engaged as stimuli become familiar over days and are suppressed by novel stimuli, in striking opposition to PV+ neuronal activity (see figure below). Consistent with a current understanding that SST+ neurons inhibit PV+ neurons in layer 4, we propose that long-term habituation is mediated by the influence of memory on SST+ neurons, which both inhibit excitatory neuronal targets through dendritic inhibition but also disinhibit those same targets by inhibiting PV+ neurons. Indeed, optogenetic stimulation of SST+ inhibitory neurons in V1 inhibits PV+ neurons to produce net disinhibition of cortical neurons. Thus, we are gaining a deep understanding of unexpected but critical circuit level interactions that support a foundational form of learning and memory.
[Picture]
Parvalbumin-expressing (PV+) GABAergic cortical neurons are engaged by novel stimuli but suppressed by familiar, while Somatostatin-expressing (SOM+) neurons show an opposing response, suggesting disinhibition. Figure from Montgomery et al. 2022
Sleep dependency
While sleep has clearly evolved with a primarily metabolic function, the consolidation of long-term memory also requires sleep. This consolidation process may involve key events within the neurons that are being modified during learning, such as the synthesis of critical proteins that alter synaptic strength in a lasting way. There may also be further activity-dependent events that replay the patterns of activity that occurred during learning, provide further reinforcement to temporary changes that occurred while the animal is awake or suppress irrelevant inputs to increase signal-to-noise. Moreover, there may be systems-level consolidation processes in which interactions between different regions of the brain are required off-line, in order to distribute information through the brain in a manner that makes memory more robust through increased reduncancy and therefore longer lasting. Despite all of these interesting theories, our understanding of the role of sleep in memory remains poor. Stimulus-selective response plasticity (SRP) provides a potential window into the involvement of sleep in learning and memory as SRP does not manifest during stimulus presentation, but only emerges after a period of sleep (see figure below). Moreover, studies have shown that disruption of sleep prevents SRP while presentation of the stimulus just prior to the animal falling asleep promotes the occurrence and magnitude of SRP. Thus, we believe that phenomena like long-term habituation and SRP provide robust and reliable assays to determine the key events occurring during sleep that contribute to memory consolidation.
While sleep has clearly evolved with a primarily metabolic function, the consolidation of long-term memory also requires sleep. This consolidation process may involve key events within the neurons that are being modified during learning, such as the synthesis of critical proteins that alter synaptic strength in a lasting way. There may also be further activity-dependent events that replay the patterns of activity that occurred during learning, provide further reinforcement to temporary changes that occurred while the animal is awake or suppress irrelevant inputs to increase signal-to-noise. Moreover, there may be systems-level consolidation processes in which interactions between different regions of the brain are required off-line, in order to distribute information through the brain in a manner that makes memory more robust through increased reduncancy and therefore longer lasting. Despite all of these interesting theories, our understanding of the role of sleep in memory remains poor. Stimulus-selective response plasticity (SRP) provides a potential window into the involvement of sleep in learning and memory as SRP does not manifest during stimulus presentation, but only emerges after a period of sleep (see figure below). Moreover, studies have shown that disruption of sleep prevents SRP while presentation of the stimulus just prior to the animal falling asleep promotes the occurrence and magnitude of SRP. Thus, we believe that phenomena like long-term habituation and SRP provide robust and reliable assays to determine the key events occurring during sleep that contribute to memory consolidation.
[Picture]
Long-term savings for familiarity, manifested as habituation, occur during sleep. This is accompanied by stimulus-selective response plasticity, which only manifests as a potentiation of the cortical response after sleep in mice.
Adaptive Filtration or Memory?
We now realise that habituation across different timescales reflects the operation of very different circuits and mechanisms, even though the net output is a reduction in behavioural response to stimuli. Currently, we are working on differentiating processes of adaptive filtration from a true memory comparator system. During adaptive filtration, synaptic connections responsive to innocuous stimuli are weakened by repeated use over short timescales. Novelty detection therefore requires elicitation of behaviour via a different neural pathway. Alternatively, memories formed in the cerebral cortex could trigger inhibitory systems to selectively suppress behavioural output across longer timescales when an innocuous sensory input matches a stored memory of that input. Under these circumstances, novelty results in behaviour when this comparison with memory produces a mismatch (see figure below). This latter system, which is described as a comparator, retains attractive properties for recognition systems, including pattern completion and gating based on context or other conditional factors. We believe that understanding the diverse mechanisms by which the brain mediates habituation and novelty detection over different timescales will allow us to gain a more general insight into how the brain learns, consolidates and retrieves memory.
We now realise that habituation across different timescales reflects the operation of very different circuits and mechanisms, even though the net output is a reduction in behavioural response to stimuli. Currently, we are working on differentiating processes of adaptive filtration from a true memory comparator system. During adaptive filtration, synaptic connections responsive to innocuous stimuli are weakened by repeated use over short timescales. Novelty detection therefore requires elicitation of behaviour via a different neural pathway. Alternatively, memories formed in the cerebral cortex could trigger inhibitory systems to selectively suppress behavioural output across longer timescales when an innocuous sensory input matches a stored memory of that input. Under these circumstances, novelty results in behaviour when this comparison with memory produces a mismatch (see figure below). This latter system, which is described as a comparator, retains attractive properties for recognition systems, including pattern completion and gating based on context or other conditional factors. We believe that understanding the diverse mechanisms by which the brain mediates habituation and novelty detection over different timescales will allow us to gain a more general insight into how the brain learns, consolidates and retrieves memory.
[Picture]
Adaptive filtration leads to a gradual loss of feedforward influence from sensory experience whereas comparator systems compare sensory experience with memory to suppress behaviour through inhibition when the stimulus is known to be innocuous
The ongoing work of our laboratory on fundamental mechanisms of habituation and novelty detection is investigating the rich variety of phenomena occurring over different timescales at the synaptic, circuit and systems levels using a wide array of both observational and interventional approaches. We believe that understanding these foundational and highly conserved, yet relatively tractable but underexplored forms of learning and memory, is likely to provide more generalized insight into how information is stored in and retrieved from the brain.