Cosyne 2009 Workshops
March 2-3, 2009
Snow Bird, Utah
Workshop Title
The role(s) of inhibition and excitatory/inhibitory balance in sensory processing
Organizer(s)
Dan Butts, University of Maryland, College Park
Abstract
Inhibition is a ubiquitous feature of neural circuits, and is clearly involved at every level of sensory systems. What is not clear, however, is exactly what role(s) inhibition plays in sensory processing.
As more direct experimental evidence of inhibition accumulates, it is becoming increasingly possible to link roles of inhibition hypothesized from system-level function with experimental observations. Such roles include gain control, generation of precise timing, shaping development and learning, preventing runaway excitation, and allowing for other system-level computation such as "winner-take-all". This workshop will bring together experimentalists and theorists studying a variety of sensory areas, with the goal of connecting single-neuron observations with system-wide computation.
Speakers
MORNING SESSION
8:00 - 8:25 am: Dan Butts (University of Maryland, College Park)
Interplay of excitation and inhibition to generate timing precision
8:30 - 8:55 am: Judith Hirsch (USC)
Inhibitory circuits in the visual thalamus
9:00 - 9:25 am: Nicholas Priebe (University of Texas, Austin)
The role of inhibition and spike threshold in stimulus selectivity
9:30 - 10:00 am: Coffee break and optional discussion
10:00 - 10:25 am: Mike Wehr (OHSU)
The role of inhibition in auditory cortex
10:30 - 10:55 am: Li Zhang (USC)
Spectrotemporal shaping of auditory responses by cortical inhibition
AFTERNOON SESSION
4:30 - 4:55 pm: Shawn Olsen (Harvard Medical School)
Inhibition, gain control, and decorrelation in the Drosophila olfactory system
5:00 - 5:25 pm: Michael Higley (Harvard Medical School)
Dynamics of sensory-evoked excitation and inhibition in cortical circuits
5:30 - 6:00 pm: Coffee break and optional discussion
6:00 - 6:25 pm: Jason Kerr (Max Planck Institute for Biological Cybernetics)
Would the real inhibitor please stand up: imaging neural populations in vivo
6:30 - 6:55 pm: Ilan Lampl (Weizmann Institute of Science)
Dynamics of excitation and inhibition during adaptation and spontaneous activities
7:00 - 7:25 pm: Patrick Kanold (University of Maryland, College Park)
Inhibition, synchrony, and cortical development
Speaker Abstracts
Dan Butts (University of Maryland, College Park): Interplay of excitation and inhibition to generate timing precision
LGN neurons can respond to visual stimuli with millisecond precision, despite the much slower generating signals in the visual pathway. Using novel non-linear modeling methods, we demonstrate that these precise responses arise through the interplay of excitation and similarly-tuned-but-delayed inhibition. The models’ success at reproducing the temporal details of LGN responses highlights the new method's ability to separately identify excitatory and inhibitory influences using only extracellular data. We explore how these separate influences are influenced by contrast, and are inherited in the retina versus generated at the level of the LGN. Together, this demonstrates how responses with rich temporal structure can arise from slower visual signals, suggesting a general role for local inhibitory circuitry in temporal processing.
Judith Hirsch (University of Southern California): Inhibitory circuits in the visual thalamus
The intrinsic circuitry of the lateral geniculate nucleus of thalamus is dominated by feedforward inhibitory circuits. Thalamic interneurons not only receive massive input from the retina but also form dense synaptic connections with relay cells and each other. By contrast, relay cells form sparse, if any, connections with their neighbors. Yet, there is scant information about how thalamic inhibition operates in vivo. To explore this subject, we make whole-cell recordings from the primary layers of cat's lateral geniculate nucleus and record visual responses from anatomically labeled neurons. Our previous studies of relay cells showed that stimuli of the reverse contrast evoke responses of the opposite sign in both the center and surround of the receptive field. This pull-pull arrangement of synaptic responses is most easily explained by direct feedforward input from the retina and indirect feedforward inhibition mediated by local interneurons. Consistent with this scheme, we find that local interneurons have receptive fields with a center-surround organization. Moreover, the receptive fields of relay cells and interneurons are not only qualitatively but also quantitatively alike; a simple model predicts the visual responses of both types of cells with equal fidelity. Given this similarity, it was surprising to observe that the shapes of the intracellular waveforms recorded from each type of neuron are almost polar opposites. Excitatory stimuli elicit rapid sequences of unitary depolarizations from relay cells but relatively smooth and graded depolarizations from interneurons. By contrast, suppressive stimuli evoke smooth and graded inhibition from relay cells but rapid trains of unitary inhibitory events from interneurons. Thus, even though relay cells and local interneurons have receptive fields with similar shapes, these two types of neurons process retinal inputs in profoundly different ways.
Nicholas Priebe (University of Texas at Austin): The role of inhibition and spike threshold in stimulus selectivity
The question of how precise selectivity emerges in the visual cortex has been marked by considerable controversy, ever since Hubel and Wiesel first described orientation selectivity. There are essentially two different views of how selectivity arises. Feed-forward models, which derive from Hubel and Wiesel’s original proposals, rely entirely on the properties and organization of thalamic or feed-forward inputs to cortical cells. Feed-back models require lateral inhibition of some form to refine response selectivity relative to the rather weak bias provided by thalamic inputs.
This debate has in large part been driven by the paradox presented by two divergent lines of evidence. On the one hand, many cortical response properties, such as cross-orientation suppression, orientation, direction and temporal frequency selectivity, appear to require lateral inhibition. On the other hand, while lateral inhibition could potentially provide considerable computational power to neuronal circuits, several lines of evidence suggest that it may not sharpen selectivity in the cerebral cortex. In intracellular recordings from primary sensory cortical areas, synaptic activity often lacks the necessary properties to support lateral inhibition. Inhibitory inputs are most often tuned to the same stimuli as the excitatory inputs, and inhibition evoked by non-preferred stimuli is generally weak. Inactivation of the cortical circuit, including both excitatory and inhibitory components, does not degrade the selectivity of the remaining feed-forward synaptic inputs.
I will outline a simple feed-forward model of cortical function that can replicate the response properties of cortical neurons in great detail without the inclusion of lateral inhibition. The complex aspects of cortical responses that have most often been attributed to lateral inhibition can be explained parsimoniously from simple, well-characterized, nonlinear features of the feed-forward excitatory pathways, such as spike threshold, contrast saturation and spike rectification.
Mike Wehr (University of Oregon): The role of inhibition in auditory cortex
Despite recent progress, the role of synaptic inhibition in auditory cortex remains enigmatic. Inhibition appears to be more involved with timing than with tuning, but much of what we have learned from in vivo whole cell studies has been about what inhibition does not do. I argue that this leaves the fundamental roles of inhibition an open question. Here I will review what we do and do not know about inhibition in auditory cortex, compare these findings to those in visual cortex, and present some recent work on the role of synaptic inhibition in receptive field plasticity in auditory cortex.
Li Zhang (University of Southern California): Spectrotemporal shaping of auditory responses by cortical inhibition
To depict the role of synaptic inhibition in shaping cortical representation of sound frequency, two conflicting models have been proposed. In one model, inhibitory input sharpens frequency tuning by lowering the spike threshold, through a potential "iceberg" effect, as suggested by the overlapping excitatory and inhibitory synaptic tonal receptive fields (TRFs) of cortical pyramidal neurons and the balanced excitation and inhibition. In the second model, frequency tuning is sharpened by pure lateral inhibition, as suggested by previous extracellular recording experiments of two-tone suppression. To resolve the conflict, we examined the detailed frequency-tuning patterns of both excitatory and inhibitory synaptic inputs in the same pyramidal neurons. Our results demonstrate that the frequency tuning curve of inhibitory input is often characterized by a plateau-like peak, and is less selective than that of excitatory input. This results in relatively stronger inhibition in frequency ranges flanking the preferred frequencies of the cell and a significant sharpening of the frequency tuning of membrane responses. These results indicate that the general suppression and an analogous lateral sharpening coexist for cortical inhibitory effect. Our current model thus unites the two previous models. In addition, by identifying cell type according to the spike property and morphological characteristics, we demonstrate that L4 fast-spiking GABAergic neurons (the major source of local inhibitory input) exhibit broader spike TRFs than nearby pyramidal neurons, although their synaptic TRFs are of similar sizes. Thus fast-spiking neurons can convert a larger range of synaptic inputs into spike outputs, contributing to the broader tuning of inhibitory input to cortical cells.
In another study, we investigated the synaptic mechanism for the cortical representation of sound intensity. Using in vivo whole-cell voltage-clamp recordings, we discovered that intensity-selective neurons receive unbalanced excitatory and inhibitory inputs evoked by the same tone stimulus at various intensity levels. The excitatory inputs change nonmonotonically with the increase of intensity, while inhibitory inputs change monotonically. In addition, the temporal delay between tone-evoked excitatory and inhibitory inputs decreases with the increased intensity, resulting in a stronger suppression of neuronal excitation by cortical inhibition, which leads to enhanced intensity tuning. Thus, cortical intensity selectivity is primarily determined by excitatory inputs and shaped by cortical inhibition through a dynamic control of excitatory and inhibitory timing.
Shawn Olsen (Harvard Medical School): Inhibition, gain control, and decorrelation in the Drosophila olfactory system
A striking feature of the early olfactory system is its segregation into many parallel processing channels represented by glomeruli. Glomeruli are interconnected by local inhibitory neurons, which have been proposed to sharpen tuning curves and coordinate spike timing. In this presentation I will discuss our investigations into the role of inhibition in the Drosophila olfactory system. We make in vivo recordings from genetically-identified neurons in combination with precise circuit manipulations. These studies reveal that local inhibitory neurons pool activity across most glomeruli and provide an inhibitory signal proportional to the net glomerular activation. We use an experimentally constrained model to explore the consequences of this inhibition for odor coding and find that global inhibition decorrelates glomerular responses.
Michael J. Higley (Harvard Medical School): Dynamics of sensory-evoked excitation and inhibition in cortical circuits
The response of cortical neurons to sensory activation is shaped by the integration of excitatory and inhibitory synaptic inputs. In the barrel subregion of primary somatosensory cortex, a single whisker deflection evokes a characteristic pattern of excitation rapidly followed by inhibition that curtails both membrane depolarization and spike output. The summation of these opposing drives contributes to the generation of receptive field properties and regulates the magnitude and precision of spike output.
Importantly, the amplitude and timing of excitatory and inhibitory inputs are not fixed, but instead exhibit strong dependence on the spatiotemporal properties of the sensory stimulus, as illustrated by a number of examples. First, the cortical response to whisker deflection is strongly reduced by preceding deflection of a neighboring whisker. This cross-whisker suppression arises largely from subcortical mechanisms that diminish thalamocortical drive, producing similar reduction in both excitation and inhibition. Second, barrel cortex neurons are highly sensitive to the direction of whisker deflection. Both excitation and inhibition are similarly enhanced for preferred versus non-preferred directions. However, the relative excitatory synaptic latency is decreased for the preferred direction, producing a greater window of net excitation and a sharpening of overall tuning. Finally, repetitive whisker deflection, such as occurs during bouts of whisking, produces response adaptation in cortical neurons. However, within layer 4, a balanced reduction in both excitatory and inhibitory synaptic conductances occurs, preserving the high precision of spike output despite a decrease in response magnitude.
Taken together, our findings reinforce the view that excitatory and inhibitory dynamics are a key determinant of cortical responses to sensory stimuli and are strongly influenced by stimulus properties. The finding that changes in stimulus parameters often produce similar, if not perfectly balanced, changes in the magnitude of excitation and inhibition suggests that cortical circuits are highly tuned to exhibit a certain degree of functional homeostasis. Understanding how local circuit architecture may promote such homeostasis and how it might be altered by physiological or pathological changes in brain state remains a key question for additional research.
Jason Kerr (Max Planck Institute for Biological Cybernetics): Would the real inhibitor please stand up: imaging neural populations in vivo
The role of inhibitory neurons embedded in neuronal circuits during spontaneous and evoked activity is far from being understood as their basic properties, such as spontaneous firing and evoked response rates, are not well characterized. I will present findings from our lab that addresses the role of identified inhibitory and excitatory neurons during sensory stimulation and discuss the implications of their properties on sensory processing within a cortical column.
Ilan Lampl (Weizmann Institute of Science): Dynamics of excitation and inhibition during adaptation and spontaneous activities
Inhibitory cells constitute only a small fraction of the population of cortical neurons, yet they are central for most network activities. The relation between excitation and inhibition, however, is far from being understood. In recent years an in vivo intracellular recording method which allows to estimate the amount of excitation and inhibition evoked by sensory stimulation was developed. Here, we have used it to investigate the balance between excitation and inhibition during the course of adaptation induced by repetitive stimulation of the vibrissa. Our study demonstrated that the balance depends on the history and frequency of whisker stimulation. In another study, we explored a long lasting controversy regarding the instantaneous relations between excitation and inhibition. While some studies suggested that these opposing inputs balance each other only on average, when measured at time scales of many milliseconds, others proposed a correlation on the fine millisecond scale. To this end we developed a new approach, based on simultaneous intracellular recordings from pairs of nearby cortical cells. Our results suggest that inhibitory control of excitation has a high temporal precision.
Patrick Kanold (University of Maryland, College Park): Inhibition, synchrony, and cortical development
Experience of the world in early life influences the development of brain function. Inhibitory circuits play a crucial role during experience dependent plasticity in the mammalian cortex. One function of inhibitory circuits might be to control precise temporal correlations between thalamic and cortical activity allowing for synaptic strengthening or weakening using spike-time dependent plasticity rules. An important feature of inhibitory circuits is the electrical synapses commonly established among their inhibitory interneurons promoting their synchronous firing. We found that elimination of electrical synapses formed by connexin36 (Cx36) reduced inhibitory efficacy by abolishing inhibitory synchrony. We found that such reduced inhibitory efficacy resulted in impaired ocular dominance (OD) plasticity in the visual cortex. OD plasticity in the absence of Cx36 is re-established by pharmacological enhancement of inhibitory transmission. Thus synchronization of inhibitory networks through electrical synapses is one key mechanism to adjust inhibitory strength, and as such inhibitory synchrony thereby regulates experience-dependent plasticity. A modulation of the strength of inhibition by synchrony has distinct advantages. The overall strength of inhibition is quickly modified depending on the size of the recruited coupled network and this complements other modulating mechanisms such as altered levels of receptor expression (by regulation of transcription, translation or insertion), which operate on a timescale of minutes to days. Thus experience-dependent plasticity is the result of the complementary interactions between neuronal networks, coupled chemically but also electrically.
