Past events

Firing rates of neurons cannot explain how we compose complex mental representations from more primitive elements. If spike time matters compositionality can easily be explained. This can easily be achieved by synfire chains. We provide indirect evidence that monkey scribbling is generated by synfire chains. Furthermore, we show by simulations that synfire chains in two distinct areas with a few random connections may learn to resonate with each other. We also show how many representations of mental elements may reside in the same small area, when practically all neurons participate in all the presentations, and yet what is represented can be identified in a few ms. In simulations, the global activity may oscillate in the gamma range without any oscillatory activity of individual neurons. When the activity of synfire chains in the two regions are bound the oscillations synchronize. We illustrate such processes in MEG recordings.

Classically, the cortex has been studied using electrophysiological techniques, which extract single-cell response profiles with great accuracy but leave other aspects of network responses largely inaccessible. Recently, in vivo two-photon calcium imaging (2PCI), has offered a new “look” into the cortex; allowing the imaging of response profiles and network dynamics from dozens of singly identified neurons simultaneously. I will present our work using both in vivo electrophysiology as well as 2PCI in the primary auditory cortex (A1) of mice highlighting the strengths and weaknesses of both. We first mapped the functional architecture of A1 in response to pure tones using 2PCI. This new “look” at A1 revealed a surprisingly high level of functional heterogeneity (measured as signal correlation vs. distance) in the face of the known tonotopic organization. The high variance of signal correlations suggested that neurons in A1 are organized in small cortical subnetworks. Additionally, I will discuss our preliminary analysis of population activity (i.e. pairwise noise correlations) and its potential for studying network dynamics in the future. Next, using in vivo loose patch clamp recordings, we studied the responses to natural sounds in a natural context – the mother-pup bond. We discovered that neuronal activation patterns to pup vocalizations are modulated by pup body odors. Specifically, pup odors significantly enhanced the responsiveness to natural calls of over a third of auditory responsive neurons in lactating females. This plasticity was absent in virgins and decreased in mothers following weaning of their pups. These experiments reveal a previously unknown interaction between natural sounds and smells in the neocortex which is context-dependent and ethologically relevant.

Abstract: For several years we have been studying the performance of rats in an odor mixture categorization task, in which an animal makes a left/right spatial choice instructed by the dominant component of a binary odor mixture. In order to better understand the neural basis of such odor guided decisions we have recorded ensembles of tens of neurons in several different brain regions during the performance of this task. I will present findings from these studies, emphasizing the nature of neural representations in the primary olfactory cortex as well as two downstream structures, orbitofrontal cortex and superior colliculus. My talk will emphasize the read-out and evaluation of sensory information by higher order brain regions and the contributions of non sensory variables to the performance of perceptual tasks.

Sleep and circadian rhythms are functionally important in all vertebrates and sleep disorders affect millions of people worldwide. While we understand that the timing and quality of sleep are regulated by circadian and homeostatic processes, the function of sleep is still enigmatic. Increasing evidence points to a role for sleep in maintaining “synaptic homeostasis”. This hypothesis suggests increases in global synaptic strength during wakefulness followed by a decrease during sleep, primarily in memory-related circuits. Hypocretins/orexins (HCRT) are neuropeptides that are important sleep-wake regulators and HCRT deficiency causes narcolepsy in humans and mammalian models. We have functionally characterized the HCRT system in zebrafish, a diurnal transparent vertebrate that is ideally suited to study neuronal anatomy along with sleep and circadian rhythms in vivo. We use time-lapse two-photon imaging in living zebrafish of pre- and post-synaptic markers to determine the dynamics of synaptic modifications during day and night and after manipulation of candidate genes. Video-tracking systems are used to monitor activity and sleep in order to link changes in gene expression and synaptic plasticity with behavioral output. We have found a functional HCRT neurons-pineal gland circuit that is able to modulate melatonin production and sleep consolidation. Importantly, we observed clock-controlled rhythmic variation in synapse number in HCRT axons projecting to the pineal gland. Furthermore, we cloned NPTX2b (neuronal activity-regulated pentraxin, NARP), a protein implicated in AMPA receptor clustering, and showed that it is a clock-controlled gene that regulates rhythmic synaptic plasticity in HCRT axons as well as the sleep promoting effect of melatonin. These data provide real-time, in vivo evidence of circadian regulation of structural synaptic plasticity. Building on this experimental approach, we developed several transgenic lines expressing a variety of excitatory and inhibitory synaptic markers and neuronal activity tools using the GAL4-UAS system. This opens the possibility of studying synaptic plasticity in other circuits, such as those involved in memory formation and learning, which are known to be sleep-dependent in mammals. Such an approach offers the opportunity to study synaptic plasticity in response to pharmacological and behavioral challenges or after genetic manipulation of key synaptic proteins, with complementary monitoring of the resulting behavior in a living vertebrate.