Population coding and behavioral choice

Oscillations in an artificial neural network convert competing inputs into a temporal code

The field of computer vision has long drawn inspiration from neuroscientific studies of the human and non-human primate visual system. The development of convolutional neural networks (CNNs), for example, was informed by the properties of simple and complex cells in early visual cortex. However, the computational relevance of oscillatory dynamics experimentally observed in the visual system are typically not considered in artificial neural networks (ANNs). Computational models of neocortical dynamics, on the other hand, rarely take inspiration from computer vision. Here, we combine methods from computational neuroscience and machine learning to implement multiplexing in a simple ANN using oscillatory dynamics. We first trained the network to classify individually presented letters. Post-training, we added temporal dynamics to the hidden layer, introducing refraction in the hidden units as well as pulsed inhibition mimicking neuronal alpha oscillations. Without these dynamics, the trained network correctly classified individual letters but produced a mixed output when presented with two letters simultaneously, indicating a bottleneck problem. When introducing refraction and oscillatory inhibition, the output nodes corresponding to the two stimuli activate sequentially, ordered along the phase of the inhibitory oscillations. Our model implements the idea that inhibitory oscillations segregate competing inputs in time. The results of our simulations pave the way for applications in deeper network architectures and more complicated machine learning problems.

Neural circuit regulation by identified modulatory projection neurons

Rhythmic behaviors (e.g., walking, breathing, and chewing) are produced by central pattern generator (CPG) circuits. These circuits are highly dynamic due to a multitude of input they receive from hormones, sensory neurons, and modulatory projection neurons. Such inputs not only turn CPG circuits on and off, but they adjust their synaptic and cellular properties to select behaviorally relevant outputs that last from seconds to hours. Similar to the contributions of fully identified connectomes to establishing general principles of circuit function and flexibility, identified modulatory neurons have enabled key insights into neural circuit modulation. For instance, while bath-applying neuromodulators continues to be an important approach to studying neural circuit modulation, this approach does not always mimic the neural circuit response to neuronal release of the same modulator. There is additional complexity in the actions of neuronally-released modulators due to: (1) the prevalence of co-transmitters, (2) local- and long-distance feedback regulating the timing of (co-)release, and (3) differential regulation of co-transmitter release. Identifying the physiological stimuli (e.g., identified sensory neurons) that activate modulatory projection neurons has demonstrated multiple “modulatory codes” for selecting particular circuit outputs. In some cases, population coding occurs, and in others circuit output is determined by the firing pattern and rate of the modulatory projection neurons. The ability to perform electrophysiological recordings and manipulations of small populations of identified neurons at multiple levels of rhythmic motor systems remains an important approach for determining the cellular and synaptic mechanisms underlying the rapid adaptability of rhythmic neural circuits.

Neural signature of the perceptual decision in the neural population responses of the inferior temporal cortex

Rapid categorization of visual objects is critical for comprehending our complex visual world. The role of individual cortical neurons and neural populations in categorizing visual objects during passive vision has previously been studied. However, it is unclear whether and how perceptually guided behaviors affect the encoding of stimulus categories by neural population activity in the higher visual cortex. Here we studied the activity of the inferior temporal (IT) cortical neurons in macaque monkeys during both passive viewing and categorization of ambiguous body and object images. We found enhanced category information in the IT neural population activity during the correct, but not wrong, trials of the categorization task compared to the passive task. This encoding enhancement was task difficulty dependent with progressively larger values in trials with more ambiguous stimuli. Enhancement of IT neural population information for behaviorally relevant stimulus features suggests IT neural networks' involvement in perceptual decision-making behavior.

Descending neurons coordinate anterior grooming behavior in Drosophila

The brain coordinates the movements that constitute behavior, but how descending neurons convey the myriad of commands required to activate the motor neurons of the limbs in the right order and combinations to produce those movements is not well understood. For anterior grooming behavior in the fly, we show that its component head sweeps and leg rubs can be initiated separately, or as a set, by different descending neurons. Head sweeps and leg rubs are mutually exclusive movements of the front legs that normally alternate, and we show that circuits in the ventral nerve cord as well as in the brain can resolve competing commands. Finally, the left and right legs must work together to remove debris. The coordination for leg rubs can be achieved by unilateral activation of a single descending neuron, while a similar manipulation of a different descending neuron decouples the legs to produce single-sided head sweeps. Taken together, these results demonstrate that distinct descending neurons orchestrate the complex alternation between the movements that make up anterior grooming.

The Complement of Projection Neurons Activated Determines the Type of Feeding Motor Program in Aplysia

Multiple projection neurons are often activated to initiate behavior. A question that then arises is, what is the unique functional role of each neuron activated? We address this issue in the feeding system of Aplysia. Previous experiments identified a projection neuron [cerebral buccal interneuron 2 (CBI-2)] that can trigger ingestive motor programs but only after it is repeatedly stimulated, i.e., initial programs are poorly defined. As CBI-2 stimulation continues, programs become progressively more ingestive (repetition priming occurs). This priming results, at least in part, from persistent actions of peptide cotransmitters released from CBI-2. We now show that in some preparations repetition priming does not occur. There is no clear seasonal effect; priming and non-priming preparations are encountered throughout the year. CBI-2 is electrically coupled to a second projection neuron, cerebral buccal interneuron 3 (CBI-3). In preparations in which priming does not occur, we show that ingestive activity is generated when CBI-2 and CBI-3 are coactivated. Programs are immediately ingestive, i.e., priming is not necessary, and a persistent state is not induced. Our data suggest that dynamic changes in the configuration of activity can vary and be determined by the complement of projection neurons that trigger activity.

Distributed control of motor circuits for backward walking in Drosophila

How do descending inputs from the brain control leg motor circuits to change how an animal walks? Conceptually, descending neurons are thought to function either as command-type neurons, in which a single type of descending neuron exerts a high-level control to elicit a coordinated change in motor output, or through a population coding mechanism, whereby a group of neurons, each with local effects, act in combination to elicit a global motor response. The Drosophila Moonwalker Descending Neurons (MDNs), which alter leg motor circuit dynamics so that the fly walks backwards, exemplify the command-type mechanism. Here, we identify several dozen MDN target neurons within the leg motor circuits, and show that two of them mediate distinct and highly-specific changes in leg muscle activity during backward walking: LBL40 neurons provide the hindleg power stroke during stance phase; LUL130 neurons lift the legs at the end of stance to initiate swing. Through these two effector neurons, MDN directly controls both the stance and swing phases of the backward stepping cycle. These findings suggest that command-type descending neurons can also operate through the distributed control of local motor circuits.

Information Closure Theory of Consciousness

Information processing in neural systems can be described and analyzed at multiple spatiotemporal scales. Generally, information at lower levels is more fine-grained but can be coarse-grained at higher levels. However, only information processed at specific scales of coarse-graining appears to be available for conscious awareness. We do not have direct experience of information available at the scale of individual neurons, which is noisy and highly stochastic. Neither do we have experience of more macro-scale interactions, such as interpersonal communications. Neurophysiological evidence suggests that conscious experiences co-vary with information encoded in coarse-grained neural states such as the firing pattern of a population of neurons. In this article, we introduce a new informational theory of consciousness: Information Closure Theory of Consciousness (ICT). We hypothesize that conscious processes are processes which form non-trivial informational closure (NTIC) with respect to the environment at certain coarse-grained scales. This hypothesis implies that conscious experience is confined due to informational closure from conscious processing to other coarse-grained scales. Information Closure Theory of Consciousness (ICT) proposes new quantitative definitions of both conscious content and conscious level. With the parsimonious definitions and a hypothesize, ICT provides explanations and predictions of various phenomena associated with consciousness. The implications of ICT naturally reconcile issues in many existing theories of consciousness and provides explanations for many of our intuitions about consciousness. Most importantly, ICT demonstrates that information can be the common language between consciousness and physical reality.

Simulating Small Neural Circuits with a Discrete Computational Model

Simulations of neural activity are commonly based on differential equations. We address the question what can be achieved with a simplified discrete model. The proposed model resembles artificial neural networks enriched with additional biologically inspired features. A neuron has several states, and the state transitions follow endogenous patterns which roughly correspond to firing behavior observed in biological neurons: oscillatory, tonic, plateauing, etc. Neural interactions consist of two components: synaptic connections and extrasynaptic emission of neurotransmitters. The dynamics is asynchronous and event-based; the events correspond to the changes in neurons activity. This model is innovative in introducing discrete framework for modeling neurotransmitter interactions which play the important role in neuromodulation. We simulate rhythmic activity of small neural ensembles like central pattern generators (CPG). The modeled examples include: the biphasic rhythm generated by the half-center mechanism with the post-inhibitory rebound (like the leech heartbeat CPG), the triphasic rhythm (like in pond snail feeding CPG) and the pattern switch in the system of several neurons (like the switch between ingestion and egestion in Aplysia feeding CPG). The asynchronous dynamics allows to obtain multi-phasic rhythms with phase durations close to their biological prototypes. The perspectives of discrete modeling in biological research are discussed in the conclusion.

A spatially dynamic network underlies the generation of inspiratory behaviors

Breathing is a vital rhythmic behavior that originates from neural networks within the brainstem. It is hypothesized that the breathing rhythm is generated by spatially distinct networks localized to discrete kernels or compartments. Here, we provide evidence that the functional boundaries of these compartments expand and contract dynamically based on behavioral or physiological demands. The ability of these rhythmic networks to change in size may allow the breathing rhythm to be very reliable, yet flexible enough to accommodate the large repertoire of mammalian behaviors that must be integrated with breathing.

The ability of neuronal networks to reconfigure is a key property underlying behavioral flexibility. Networks with recurrent topology are particularly prone to reconfiguration through changes in synaptic and intrinsic properties. Here, we explore spatial reconfiguration in the reticular networks of the medulla that generate breathing. Combined results from in vitro and in vivo approaches demonstrate that the network architecture underlying generation of the inspiratory phase of breathing is not static but can be spatially redistributed by shifts in the balance of excitatory and inhibitory network influences. These shifts in excitation/inhibition allow the size of the active network to expand and contract along a rostrocaudal medullary column during behavioral or metabolic challenges to breathing, such as changes in sensory feedback, sighing, and gasping. We postulate that the ability of this rhythm-generating network to spatially reconfigure contributes to the remarkable robustness and flexibility of breathing.

Multifunctional Wing Motor Control of Song and Flight

Multifunctional motor systems produce distinct output patterns that are dependent on behavioral context, posing a challenge to underlying neuronal control. Flies use their wings for flight and the production of a patterned acoustic signal, the male courtship song, employing in both cases a small set of wing muscles and corresponding motor neurons. We took first steps toward elucidating the neuronal control mechanisms of this multifunctional motor system by live imaging of muscle ensemble activity patterns during song and flight, and we established the functional role of a comprehensive set of wing muscle motor neurons by silencing experiments. Song and flight rely on distinct configurations of neuromuscular activity, with most, but not all, flight muscles and their corresponding motor neurons contributing to song and shaping its acoustic parameters. The two behaviors are exclusive, and the neuronal command for flight overrides the command for song. The neuromodulator octopamine is a candidate for selectively stabilizing flight, but not song motor patterns.

Robust mixture modeling reveals category-free selectivity in reward region neuronal ensembles

Classification of neurons into clusters based on their response properties is an important tool for gaining insight into neural computations. However, it remains unclear to what extent neurons fall naturally into discrete functional categories. We developed a Bayesian method that models the tuning properties of neural populations as a mixture of multiple types of task-relevant response patterns. We applied this method to data from several cortical and striatal regions in economic choice tasks. In all cases, neurons fell into only two clusters: one multiple-selectivity cluster containing all cells driven by task variables of interest and another of no selectivity for those variables. The single cluster of task-sensitive cells argues against robust categorical tuning in these areas. The no-selectivity cluster was unanticipated and raises important questions about what distinguishes these neurons and what role they play. Moreover, the ability to formally identify these nonselective cells allows for more accurate measurement of ensemble effects by excluding or appropriately down-weighting them in analysis. Our findings provide a valuable tool for analysis of neural data, challenge simple categorization schemes previously proposed for these regions, and place useful constraints on neurocomputational models of economic choice and control. NEW & NOTEWORTHY We present a Bayesian method for formally detecting whether a population of neurons can be naturally classified into clusters based on their response tuning properties. We then examine several data sets of reward system neurons for variables and find in all cases that neurons can be classified into only two categories: a functional class and a non-task-driven class. These results provide important constraints for neural models of the reward system.

Control without Controllers: Toward a Distributed Neuroscience of Executive Control

Executive control refers to the regulation of cognition and behavior by mental processes and is a hallmark of higher cognition. Most approaches to understanding its mechanisms begin with the assumption that our brains have anatomically segregated and functionally specialized control modules. The modular approach is intuitive: Control is conceptually distinct from basic mental processing, so an organization that reifies that distinction makes sense. An alternative approach sees executive control as self-organizing principles of a distributed organization. In distributed systems, control and controlled processes are colocalized within large numbers of dispersed computational agents. Control then is often an emergent consequence of simple rules governing the interaction between agents. Because these systems are unfamiliar and unintuitive, here we review several well-understood examples of distributed control systems, group living insects and social animals, and emphasize their parallels with neural systems. We then reexamine the cognitive neuroscience literature on executive control for evidence that its neural control systems may be distributed.

Using Matrix and Tensor Factorizations for the Single-Trial Analysis of Population Spike Trains

Advances in neuronal recording techniques are leading to ever larger numbers of simultaneously monitored neurons. This poses the important analytical challenge of how to capture compactly all sensory information that neural population codes carry in their spatial dimension (differences in stimulus tuning across neurons at different locations), in their temporal dimension (temporal neural response variations), or in their combination (temporally coordinated neural population firing). Here we investigate the utility of tensor factorizations of population spike trains along space and time. These factorizations decompose a dataset of single-trial population spike trains into spatial firing patterns (combinations of neurons firing together), temporal firing patterns (temporal activation of these groups of neurons) and trial-dependent activation coefficients (strength of recruitment of such neural patterns on each trial). We validated various factorization methods on simulated data and on populations of ganglion cells simultaneously recorded in the salamander retina. We found that single-trial tensor space-by-time decompositions provided low-dimensional data-robust representations of spike trains that capture efficiently both their spatial and temporal information about sensory stimuli. Tensor decompositions with orthogonality constraints were the most efficient in extracting sensory information, whereas non-negative tensor decompositions worked well even on non-independent and overlapping spike patterns, and retrieved informative firing patterns expressed by the same population in response to novel stimuli. Our method showed that populations of retinal ganglion cells carried information in their spike timing on the ten-milliseconds-scale about spatial details of natural images. This information could not be recovered from the spike counts of these cells. First-spike latencies carried the majority of information provided by the whole spike train about fine-scale image features, and supplied almost as much information about coarse natural image features as firing rates. Together, these results highlight the importance of spike timing, and particularly of first-spike latencies, in retinal coding.

Understanding Neural Population Coding: Information Theoretic Insights from the Auditory System

In recent years, our research in computational neuroscience has focused on understanding how populations of neurons encode naturalistic stimuli. In particular, we focused on how populations of neurons use the time domain to encode sensory information. In this focused review, we summarize this recent work from our laboratory. We focus in particular on the mathematical methods that we developed for the quantification of how information is encoded by populations of neurons and on how we used these methods to investigate the encoding of complex naturalistic sounds in auditory cortex. We review how these methods revealed a complementary role of low frequency oscillations and millisecond precise spike patterns in encoding complex sounds and in making these representations robust to imprecise knowledge about the timing of the external stimulus. Further, we discuss challenges in extending this work to understand how large populations of neurons encode sensory information. Overall, this previous work provides analytical tools and conceptual understanding necessary to study the principles of how neural populations reflect sensory inputs and achieve a stable representation despite many uncertainties in the environment.

Toward an organismal neurobiology: integrative neuroethology

Overt behavior is generated in response to a palette of external and internal stimuli and internal drives. Rarely are these variables introduced in isolation. This creates challenges for the organism to sort inputs that frequently favor conflicting behaviors. Under these conditions, the nervous system relies on established and flexible hierarchies to produce appropriate behavioral changes. The pteropod mollusc Clione limacina is used as an example to illustrate a variety of behavioral interactions that alter a baseline behavioral activity: slow swimming. The alterations include acceleration within the slow swimming mode, acceleration from the slow to fast swimming modes, whole body withdrawal (and inhibition of swimming), food acquisition behavior (with a feeding motivational state), and a startle locomotory response. These examples highlight different types of interaction between the baseline behavior and the new behaviors that involve external stimuli and two types of internal drives: a modular arousal system and a motivational state. The investigation of hierarchical interactions between behavioral modules is a central theme of integrative neuroethology that focuses on an organismal level of understanding of the neural control of behavior.

The dynamics of analogue signalling in local networks controlling limb movement

Communication by analogue signals is relatively common in arthropod local networks. In the locust, non-spiking local interneurons play a key role in controlling sets of motor neurons in the generation of local reflex movements of the limbs. Here, our aim was two-fold. Our first aim was to determine the coding properties of a subpopulation of these interneurons by using system identification approaches. To this end, the femoro-tibial chordotonal organ, which monitors the movements of the tibia about the femur, was stimulated with Gaussian white noise and with more natural stimuli corresponding to the movements of the tibia during walking. The results showed that the sample of interneurons analysed displayed a wide, and overlapping, range of response characteristics. The second aim was to develop and test improved data analysis methods for describing neuronal function that are more robust and allow statistical analysis, a need emphasized by the high levels of background neuronal activity usually observed. We found that nonlinear models provided an improved fit in describing the response properties of interneurons that were then classified with statistical clustering methods. We identified four distinct categories of interneuron response that can be further divided into nine groups, with most interneurons being excited during extension movements of the leg, reflecting the outputs of upstream spiking local interneurons.

Neural circuit flexibility in a small sensorimotor system

Neuronal circuits underlying rhythmic behaviors (central pattern generators: CPGs) can generate rhythmic motor output without sensory input. However, sensory input is pivotal for generating behaviorally relevant CPG output. Here we discuss recent work in the decapod crustacean stomatogastric nervous system (STNS) identifying cellular and synaptic mechanisms whereby sensory inputs select particular motor outputs from CPG circuits. This includes several examples in which sensory neurons regulate the impact of descending projection neurons on CPG circuits. This level of analysis is possible in the STNS due to the relatively unique access to identified circuit, projection, and sensory neurons. These studies are also revealing additional degrees of freedom in sensorimotor integration that underlie the extensive flexibility intrinsic to rhythmic motor systems.

Multiplexing of motor information in the discharge of a collision detecting neuron during escape behaviors

Locusts possess an identified neuron, the descending contralateral movement detector (DCMD), conveying visual information about impending collision from the brain to thoracic motor centers. We built a telemetry system to simultaneously record, in freely behaving animals, the activity of the DCMD and of motoneurons involved in jump execution. Cocontraction of antagonistic leg muscles, a required preparatory phase, was triggered after the DCMD firing rate crossed a threshold. Thereafter, the number of DCMD spikes predicted precisely motoneuron activity and jump occurrence. Additionally, the time of DCMD peak firing rate predicted that of jump. Ablation experiments suggest that the DCMD, together with a nearly identical ipsilateral descending neuron, is responsible for the timely execution of the escape. Thus, three distinct features that are multiplexed in a single neuron's sensory response to impending collision-firing rate threshold, peak firing time, and spike count-probably control three distinct motor aspects of escape behaviors.

Shared Strategies for Behavioral Switching: Understanding How Locomotor Patterns are Turned on and Off

Animals frequently switch from one behavior to another, often to meet the demands of their changing environment or internal state. What factors control these behavioral switches and the selection of what to do or what not to do? To address these issues, we will focus on the locomotor behaviors of two distantly related “worms,” the medicinal leech Hirudo verbana (clade Lophotrochozoa) and the nematode Caenorhabditis elegans (clade Ecdysozoa). Although the neural architecture and body morphology of these organisms are quite distinct, they appear to switch between different forms of locomotion by using similar strategies of decision-making. For example, information that distinguishes between liquid and more solid environments dictates whether an animal swims or crawls. In the leech, dopamine biases locomotor neural networks so that crawling is turned on and swimming is turned off. In C. elegans, dopamine may also promote crawling, a form of locomotion that has gained new attention.

Multifunctional and specialized spinal interneurons for turtle limb movements

The turtle spinal cord can help reveal how vertebrate central nervous system (CNS) circuits select and generate an appropriate limb movement in each circumstance. Both multifunctional and specialized spinal interneurons contribute to the motor patterns for the three forms of scratching, forward swimming, and flexion reflex. Multifunctional interneurons, activated during all of these motor patterns, can have axon terminal arborizations in the ventral horn, where they likely contribute to limb motor output. Specialized interneurons can be specialized for a behavior, as opposed to a phase or motor synergy. Interneurons specialized for scratching can be hyperpolarized throughout swimming. Interneurons specialized for flexion reflex can be hyperpolarized throughout scratching and swimming. Some structure-function correlations have been revealed: flexion reflex-selective interneurons had somata exclusively in the dorsal horn, in contrast to scratch-activated interneurons. Transverse interneurons, defined by quantitative morphological criteria, had higher peak firing rates, narrower action potentials, briefer afterhyperpolarizations, and larger membrane potential oscillations than scratch-activated interneurons with different dendritic morphologies. Future investigations will focus on how multifunctional and specialized spinal interneurons interact to generate each motor output.

Sensory neural codes using multiplexed temporal scales

Determining how neuronal activity represents sensory information is central for understanding perception. Recent work shows that neural responses at different timescales can encode different stimulus attributes, resulting in a temporal multiplexing of sensory information. Multiplexing increases the encoding capacity of neural responses, enables disambiguation of stimuli that cannot be discriminated at a single response timescale, and makes sensory representations stable to the presence of variability in the sensory world. Thus, as we discuss here, temporal multiplexing could be a key strategy used by the brain to form an information-rich and stable representation of the environment.

Coding characteristics of spiking local interneurons during imposed limb movements in the locust

The performance of adaptive behavior relies on continuous sensory feedback to produce relevant modifications to central motor patterns. The femoral chordotonal organ (FeCO) of the legs of the desert locust monitors the movements of the tibia about the femoro-tibial joint. A ventral midline population of spiking local interneurons in the metathoracic ganglia integrates inputs from the FeCO. We used a Wiener kernel cross-correlation method combined with a Gaussian white noise stimulation of the FeCO to completely characterize and model the output dynamics of the ventral midline population of interneurons. A wide range of responses were observed, and interneurons could be classified into three broad groups that received excitatory and inhibitory or principally inhibitory or excitatory synaptic inputs from the FeCO. Interneurons that received mixed inputs also had the greatest linear responses but primarily responded to extension of the tibia and were mostly sensitive to stimulus velocity. Interneurons that received principally inhibitory inputs were sensitive to extension and to joint position. A small group of interneurons received purely excitatory synaptic inputs and were also sensitive to tibial extension. In addition to capturing the linear and nonlinear dynamics of this population of interneurons, first- and second-order Wiener kernels revealed that the dynamics of the interneurons in the population were graded and formed a spectrum of responses whereby the activity of many cells appeared to be required to adequately describe a particular stimulus characteristic, typical of population coding.

Shared versus specialized glycinergic spinal interneurons in axial motor circuits of larval zebrafish

The neuronal networks in spinal cord can produce a diverse array of motor behaviors. In aquatic vertebrates such as fishes and tadpoles, these include escape behaviors, swimming across a range of speeds, and struggling. We addressed the question of whether these behaviors are accomplished by a shared set of spinal interneurons activated in different patterns or, instead, involve specialized spinal interneurons that may shape the motor output to produce particular behaviors. We used larval zebrafish because they are capable of several distinct axial motor behaviors using a common periphery and a relatively small set of spinal neurons, easing the task of exploring the extent to which cell types are specialized for particular motor patterns. We performed targeted in vivo whole-cell patch recordings in 3 d post fertilization larvae to reveal the activity pattern of four commissural glycinergic interneuron types during escape, swimming and struggling behaviors. While some neuronal classes were shared among different motor patterns, we found others that were active only during a single one. These specialized neurons had morphological and functional properties consistent with a role in shaping key features of the motor behavior in which they were active. Our results, in combination with other evidence from excitatory interneurons, support the idea that patterns of activity in a core network of shared spinal neurons may be shaped by more specialized interneurons to produce an assortment of motor behaviors.

State-dependent presynaptic inhibition regulates central pattern generator feedback to descending inputs

Central pattern generators (CPGs) provide feedback to their projection neuron inputs. However, it is unknown whether this feedback is regulated and how that might shape CPG output. We are studying feedback from the pyloric CPG to identified projection neurons that regulate the gastric mill CPG, in the crab stomatogastric nervous system. Both CPGs are located in the stomatogastric ganglion (STG) and are influenced by projection neurons originating in the paired commissural ganglia (CoGs). Two extrinsic inputs [ventral cardiac neurons (VCNs) and postoesophageal commissure (POC) neurons] trigger distinct gastric mill rhythms despite acting via the same projection neurons [modulatory commissural neuron 1 (MCN1); commissural projection neuron 2 (CPN2)]. These projection neurons receive feedback inhibition from the pyloric CPG interneuron anterior burster (AB), resulting in their exhibiting pyloric-timed activity during the retraction phase of the VCN- and POC-triggered gastric mill rhythms. However, during the gastric mill protraction phase, MCN1/CPN2 exhibit pyloric-timed activity during the POC-triggered rhythm but fire tonically during the VCN-triggered rhythm. Here, we show that the latter, tonic activity pattern results from the elimination of AB inhibition of MCN1/CPN2, despite persistent AB actions within the STG and AB action potentials still propagating into each CoG. This loss of pyloric-timed AB input likely results from presynaptic inhibition of AB in each CoG because, when a secondary rhythmic AB burst initiation zone in the CoG is activated, the associated action potentials are selectively suppressed during the VCN protraction phase. Thus, rhythmic CPG feedback can be locally regulated, in a state-dependent manner, enabling the same projection neurons to drive multiple motor patterns from the same neuronal circuit.

Multifunctional pattern-generating circuits

The ability of distinct anatomical circuits to generate multiple behavioral patterns is widespread among vertebrate and invertebrate species. These multifunctional neuronal circuits are the result of multistable neural dynamics and modular organization. The evidence suggests multifunctional circuits can be classified by distinct architectures, yet the activity patterns of individual neurons involved in more than one behavior can vary dramatically. Several mechanisms, including sensory input, the parallel activity of projection neurons, neuromodulation, and biomechanics, are responsible for the switching between patterns. Recent advances in both analytical and experimental tools have aided the study of these complex circuits.

Behavioral and neural responses of juvenile crayfish to moving shadows

One of the most important decisions any animal has to make is how to respond to sensory cues that suggest an imminent attack by a predator. We measured behavioral and neural responses of juvenile crayfish to moving shadows of different velocities while the animals were searching for food. In all experiments, and independent of shadow velocity, each crayfish produced one of two discrete behavioral outputs: it either tail-flipped backwards by rapid flexion of its abdomen or it immediately stopped its forward locomotion. The probability of each behavioral response was dependent on the velocity of the shadows that were presented. While most animals responded with tail-flips to slow-moving shadows and stops were rarely observed, the number of tail-flips decreased as shadow velocity increased. Tail-flips were almost absent for very fast-moving shadows and stopping behavior became the dominating response. By using a non-invasive technique to record neural activity, we were able to identify the underlying neural circuit that controlled the observed tail-flips. All tail-flips were mediated by activation of the medial giant neurons, which are part of a hardwired neural circuit previously described to produce reflexive responses to tactile stimulation.

A newly identified extrinsic input triggers a distinct gastric mill rhythm via activation of modulatory projection neurons

Neuronal network flexibility enables animals to respond appropriately to changes in their internal and external states. We are using the isolated crab stomatogastric nervous system to determine how extrinsic inputs contribute to network flexibility. The stomatogastric system includes the well-characterized gastric mill (chewing) and pyloric (filtering of chewed food) motor circuits in the stomatogastric ganglion. Projection neurons with somata in the commissural ganglia (CoGs) regulate these rhythms. Previous work characterized a unique gastric mill rhythm that occurred spontaneously in some preparations, but whose origin remained undetermined. This rhythm includes a distinct protractor phase activity pattern, during which a key gastric mill circuit neuron (LG neuron) and the projection neurons MCN1 and CPN2 fire in a pyloric rhythm-timed activity pattern instead of the tonic firing pattern exhibited by these neurons during previously studied gastric mill rhythms. Here we identify a new extrinsic input, the post-oesophageal commissure (POC) neurons, relatively brief stimulation (30 s) of which triggers a long-lasting (tens of minutes) activation of this novel gastric mill rhythm at least in part via its lasting activation of MCN1 and CPN2. Immunocytochemical and electrophysiological data suggest that the POC neurons excite MCN1 and CPN2 by release of the neuropeptide Cancer borealis tachykinin-related peptide Ia (CabTRP Ia). These data further suggest that the CoG arborization of the POC neurons comprises the previously identified anterior commissural organ (ACO), a CabTRP Ia-containing neurohemal organ. This endocrine organ thus appears to also have paracrine actions, including activation of a novel and lasting gastric mill rhythm.

Population vector code: a geometric universal as actuator

The population vector code relates directional tuning of single cells and global, directional motion incited by an assembly of neurons. In this paper three things are done. First, we analyze the population vector code as a purely geometric construct, focusing attention on its universality. Second, we generalize the algorithm on the basis of its geometrical realization so that the same construct that responds to sensation can function as an actuator for behavioral output. Third, we suggest at least a partial answer to the question of what many maps, neuronal representations of the outside sensory world in space-time, are good for: encoding vectorial input they enable a direct realization of the population vector code.

Physiology and morphology of shared and specialized spinal interneurons for locomotion and scratching

Distinct types of rhythmic movements that use the same muscles are typically generated largely by shared multifunctional neurons in invertebrates, but less is known for vertebrates. Evidence suggests that locomotion and scratching are produced partly by shared spinal cord interneuronal circuity, although direct evidence with intracellular recording has been lacking. Here, spinal interneurons were recorded intracellularly during fictive swimming and fictive scratching in vivo and filled with Neurobiotin. Some interneurons that were rhythmically activated during both swimming and scratching had axon terminal arborizations in the ventral horn of the hindlimb enlargement, indicating their likely contribution to hindlimb motor outputs during both behaviors. We previously described a morphological group of spinal interneurons ("transverse interneurons" or T neurons) that were rhythmically activated during all forms of fictive scratching at higher peak firing rates and with larger membrane potential oscillations than scratch-activated spinal interneurons with different dendritic orientations. The current study demonstrates that T neurons are activated during both swimming and scratching and thus are components of the shared circuitry. Many spinal interneurons activated during fictive scratching are also activated during fictive swimming (scratch/swim neurons), but others are suppressed during swimming (scratch-specialized neurons). The current study demonstrates that some scratch-specialized neurons receive strong and long-lasting hyperpolarizing inhibition during fictive swimming and are also morphologically distinct from T neurons. Thus this study indicates that locomotion and scratching are produced by a combination of shared and dedicated interneurons whose physiological and morphological properties are beginning to be revealed.

Modulation of locomotor activity in larval zebrafish during light adaptation

The neural basis of behavioral choice in vertebrates remains largely unknown. Zebrafish larvae have a defined locomotor repertoire as well as a simple nervous system and are therefore an attractive vertebrate system in which to study this process. Here we describe a high-throughput system for quantifying the kinematics of motor events in zebrafish larvae in order to measure the initiation frequency of different maneuvers. We use this system to analyze responses to photic stimuli and find that larvae respond to changes in illumination with both acute responses and extended behavioral programs. Reductions in illumination elicit large angle turns, distinct from startle responses, which orient larvae toward the source of light. In continuing darkness, larvae are transiently hyperactive before adopting a quiescent state. Indeed, locomotor activity is controlled by the state of light or dark adaptation similar to masking phenomena in higher vertebrates where light directly regulates motor activity. We propose that regulation of motor activity by photic stimuli in zebrafish larvae serves a behavioral goal of maximizing exposure to well lit environments optimal for feeding.

Trial-by-Trial Transformation of Error Into Sensorimotor Adaptation Changes With Environmental Dynamics

Humans can rapidly change their motor output to make goal-directed reaching movements in a new environment. Theories that describe this adaptive process have long presumed that adaptive steps scale proportionally with error. Here we show that while performing a novel reaching task, participants did not adopt a fixed learning rule, but instead modified their adaptive response based on the statistical properties of the movement environment. We found that as the directional bias of the force distribution shifted from strongly biased to unbiased, participants transitioned from an adaptive process that scaled proportionally with error to one that adapted to the direction, but not magnitude, of error. Participants also modified their response as the likelihood of the perturbation changed; as the likelihood decreased from 80 to 20% of trials, participants adopted an increasingly disproportional strategy. We propose that people can rapidly switch between learning processes within minutes of experiencing a novel environment.

Spinal interneurons that are selectively activated during fictive flexion reflex

Behavioral choices in invertebrates are mediated by a combination of shared and specialized circuitry, including neurons that are inhibited during competing behaviors. Less is known, however, about the neural mechanisms of behavioral choice in vertebrates. The spinal cord can appropriately select among several types of limb movements, including limb withdrawal (flexion reflex), scratching, and locomotion, and thus is conducive to examination of vertebrate mechanisms of behavioral choice. Flexion reflex can interrupt and reset the rhythm of scratching and locomotion, suggesting that a combination of shared and specialized circuitry contributes to these behaviors, but little is known about the interneurons involved. Here, I used in vivo intracellular recording and dye injection to identify a group of spinal interneurons that are strongly activated during fictive flexion reflex but inhibited during fictive scratching and fictive swimming. These flexion-selective interneurons are typically rhythmically hyperpolarized during fictive scratching and fictive swimming. This hyperpolarization can be maximal during the ipsilateral hip flexor bursts of rhythmic limb motor patterns, although these cells are strongly activated during the ipsilateral hip flexor bursts of fictive flexion reflex. Thus, these interneurons are relatively specialized for fictive limb withdrawal, rather than contributing to the hip flexor phase of multiple types of limb movements. These flexion-selective cells are physiologically and morphologically distinguishable from a recently described group of spinal interneurons (transverse interneurons) that are strongly activated during both fictive flexion reflex and fictive scratching. Thus, spinal interneurons with distinct behavioral roles may to some extent be morphologically distinguishable.

Multiple contributions of an input-representing neuron to the dynamics of the aplysia feeding network

In Aplysia, mutually antagonistic ingestive and egestive behaviors are produced by the same multifunctional central pattern generator (CPG) circuit. Interestingly, higher-order inputs that activate the CPG do not directly specify whether the resulting motor program is ingestive or egestive because the slow dynamics of the network intervene. One input, the commandlike cerebral-buccal interneuron 2 (CBI-2), slowly drives the motor output toward ingestion, whereas another input, the esophageal nerve (EN), drives the motor output toward egestion. When the input is switched from EN to CBI-2, the motor output does not switch immediately and remains egestive. Here, we investigated how these slow dynamics are implemented on the interneuronal level. We found that activity of two CPG interneurons, B20 and B40, tracked the motor output regardless of the input, whereas activity of another CPG interneuron, B65, tracked the input regardless of the motor output. Furthermore, we show that the slow dynamics of the network are implemented, at least in part, in the slow dynamics of the interaction between the input-representing and the output-representing neurons. We conclude that 1) a population of CPG interneurons, recruited during a particular motor program, simultaneously encodes both the input that is used to elicit the motor program and the output elicited by this input; and 2) activity of the input-representing neurons may serve to bias the future motor programs.

Imaging dedicated and multifunctional neural circuits generating distinct behaviors

Central pattern generators (CPGs) control both swimming and crawling in the medicinal leech. To investigate whether the neurons comprising these two CPGs are dedicated or multifunctional, we used voltage-sensitive dye imaging to record from approximately 80% of the approximately 400 neurons in a segmental ganglion. By eliciting swimming and crawling in the same preparation, we were able to identify neurons that participated in either of the two rhythms, or both. More than twice as many cells oscillated in-phase with crawling (188) compared with swimming (90). Surprisingly, 84 of the cells (93%) that oscillated with swimming also oscillated with crawling. We then characterized two previously unidentified interneurons, cells 255 and 257, that had interesting activity patterns based on the imaging results. Cell 255 proved to be a multifunctional interneuron that oscillates with and can perturb both rhythms, whereas cell 257 is an interneuron dedicated to crawling. These results show that the swimming and crawling networks are driven by both multifunctional and dedicated circuitry.

Optical imaging of neuronal populations during decision-making

We investigated decision-making in the leech nervous system by stimulating identical sensory inputs that sometimes elicit crawling and other times swimming. Neuronal populations were monitored with voltage-sensitive dyes after each stimulus. By quantifying the discrimination time of each neuron, we found single neurons that discriminate before the two behaviors are evident. We used principal component analysis and linear discriminant analysis to find populations of neurons that discriminated earlier than any single neuron. The analysis highlighted the neuron cell 208. Hyperpolarizing cell 208 during a stimulus biases the leech to swim; depolarizing it biases the leech to crawl or to delay swimming.

Mechanosensory activation of a motor circuit by coactivation of two projection neurons

Individual neuronal circuits can generate multiple activity patterns because of the influence of different projection neurons. However, in most systems it has been difficult to identify and assess the relative contribution of all upstream neurons responsible for the activation of any single activity pattern by a behaviorally relevant stimulus. To elucidate this issue, we used the stomatogastric nervous system (STNS) of the crab. The STNS includes the gastric mill (chewing) motor circuit in the stomatogastric ganglion (STG) and no more than 20 projection neurons that innervate the STG. We previously identified at least some (four) of the projection neurons that are activated directly by the ventral cardiac neuron (VCN) system, a population of mechanosensory neurons that activates the gastric mill circuit. Here we show that two of these projection neurons, the previously identified modulatory commissural neuron 1 (MCN1) and commissural projection neuron 2 (CPN2), are necessary and likely sufficient for the initiation/maintenance of the VCN-elicited gastric mill rhythm. Selective inactivation of either MCN1 or CPN2 still enabled a VCN-elicited gastric mill rhythm. However, because MCN1 and CPN2 have different actions on gastric mill neurons, these manipulations resulted in rhythms distinct from each other and from that occurring in the intact system. After removal of both MCN1 and CPN2, VCN stimulation failed to activate the gastric mill rhythm. Selective conjoint stimulation of MCN1 and CPN2, approximating their VCN-elicited activity patterns and firing frequencies, elicited a VCN-like gastric mill rhythm. Thus the VCN mechanosensory system elicits the gastric mill rhythm via its activation of a subset of the relevant projection neurons.

Different sensory systems share projection neurons but elicit distinct motor patterns

Considerable research has focused on issues pertaining to sensorimotor integration, but in most systems precise information remains unavailable regarding the specific pathways by which different sensory systems regulate any single central pattern-generating circuit. We address this issue by determining how two muscle stretch-sensitive neurons, the gastropyloric receptor neurons (GPRs), influence identified projection neurons that regulate the gastric mill circuit in the stomatogastric nervous system of the crab and then comparing these actions with those of the ventral cardiac neuron (VCN) mechanosensory system. Here, we show that the GPR neurons activate the gastric mill rhythm in the stomatogastric ganglion (STG) via their excitation of two identified projection neurons, modulatory commissural neuron 1 (MCN1) and commissural projection neuron 2 (CPN2), in the commissural ganglion. Support for this conclusion comes from the ability of the modulatory proctolin neuron (MPN), a projection neuron that suppresses the gastric mill rhythm via its inhibitory actions on MCN1 and CPN2, to inhibit the GPR-elicited gastric mill rhythm. Selective elimination of MCN1 and CPN2 access to the STG also prevents GPR activation of this rhythm. The VCN neurons also elicit the gastric mill rhythm by coactivating MCN1 and CPN2, but the GPR-elicited gastric mill rhythm is distinct. These distinct rhythms are likely to result partly from different MCN1 activity levels under these two conditions and partly from the presence of additional GPR actions in the STG. These results support the hypothesis that different sensory systems differentially regulate neuronal circuit activity despite their convergent actions on a single subpopulation of projection neurons.

A specific synaptic pathway activates a conditional plateau potential underlying protraction phase in the Aplysia feeding central pattern generator

A common feature in the architecture of neuronal networks is a high degree of seemingly redundant synaptic connectivity. In many cases, the synaptic inputs converging on any particular neuron all use the same neurotransmitter and appear to be fundamentally equivalent. Here, we analyze a striking counterexample in which such inputs are not equivalent and, as a result, play very different roles in the generation of the pattern of activity produced by the network. In the feeding central pattern generator of Aplysia, the pattern-initiating neuron B50 elicits motor programs by exciting the plateauing neuron B31/B32 in two ways: directly and indirectly through neuron B63. All of the synaptic connections use ACh. Despite the direct input of B50 to B31/B32, the indirect pathway of exciting B31/B32 through B63 is required for B50 to elicit the B31/B32 plateau potential and the motor program. We dissect this requirement using the muscarinic cholinergic antagonist pirenzepine. Pirenzepine blocks the B50-elicited motor program, the plateau potential in B31/B32, and, notably, a slow component of the EPSP elicited in B31/B32 by B63 but not that elicited by B50. The muscarinic agonist oxotremorine restores the plateau potential in B31/B32 and eliminates the necessity for B63 in B50-elicited motor programs. Together, our analysis shows that the plateau potential in B31/B32 is not endogenous but conditional, furthermore conditional on one particular synaptic input, that from B63. Thus, among several inputs to B31/B32 that use the same transmitter, the input from B63 is functionally distinct in its preferential access to the plateau potential that represents the committed step toward the initiation of a motor program.

Imaging reveals synaptic targets of a swim-terminating neuron in the leech CNS

In the leech, the command-like neuron called cell Tr2 is known to stop swimming, but the connections from cell Tr2 to the swim central pattern generator have not been identified. We used fluorescence resonance energy transfer voltage-sensitive dyes to identify three neurons that are synaptic targets of cell Tr2. We then used electrophysiological techniques to show that these connections are monosynaptic, chemical, and excitatory. Two of the novel targets, cell 256 and cell 54, terminate swimming when stimulated. These neurons are likely to mediate swim cessation caused by cell Tr2 activity, and thus play the role of intermediate control cells in the leech CNS.

Cerebral metabolism and consciousness

In vivo 13C magnetic resonance spectroscopy studies of the brain have measured rates of glutamate-glutamine cycle (Vcyc) and glucose oxidation (CMRglc(ox)) by detecting 13C label turnover from glucose to glutamate and glutamine. In both the awake human and in the anesthetized rat brains Vcyc and CMRglc(ox) are stoichiometrically related, and form a major pathway in which approximately 80% of the energy from glucose oxidation supports events associated with glutamate neurotransmission. The high energy consumption of the brain at rest and its quantitative usage for neurotransmission reflect a high level of neuronal activity for the non-stimulated brain. This high activity supports a reinterpretation of functional imaging data, e.g., where the large baseline signal has commonly been discarded. Independent measurements of energy consumption (delta CMRO2%) obtained from calibrated fMRI equaled percentage changes in neuronal spiking rate (delta nu %) measured by electrodes during sensory stimulation at two depths of anesthesia. These quantitative biophysical relationships between energy consumption and neuronal activity provide novel insights into the nature of brain function. The high resting brain activity is proposed to include the global interactions constituting the subjective aspects of consciousness. Anesthesia by lowering the total firing rates correlates with the loss of consciousness. These results, which measure the localized neuronal response and distinguish inputs of peripheral neurons from inputs of neurons from other brain regions, fit comfortably into the neuronal scheme of a global workspace proposed by Dehaene and Changeux.

Behavioral choices: how neuronal networks make decisions

To survive, animals must constantly make behavioral choices. The analysis of simple, almost binary, behavioral choices in invertebrate animals with restricted nervous systems is beginning to yield insight into how neuronal networks make such decisions.

Evidence for sequential decision making in the medicinal leech

Decision making can be a complex task involving a sequence of subdecisions. For example, we decide to pursue a goal (e.g., get something to eat), then decide how to accomplish that goal (e.g., go to a restaurant), and then make a sequence of more specific plans (e.g., which restaurant to go to, how to get there, what to order, etc.). In characterizing the effects of stimulating individual brain neurons in the isolated nervous system of the leech Hirudo medicinalis, we have found evidence that leeches also make decisions sequentially. In this study, we describe a pair of interneurons that elicited locomotory motor programs, either swimming or crawling, in isolated nerve cords. In semi-intact animals, stimulating the same neurons also produced either swimming or crawling, and which behavior was produced could be controlled experimentally by manipulating the depth of saline around the intact part of the leech. These same neurons were excited and fired strongly when swimming or crawling occurred spontaneously or in response to mechanosensory stimulation. We conclude that these brain interneurons help to decide on locomotion (i.e., they are "locomotory command-like neurons") and that the ultimate behavior is determined downstream, in a part of the decision-making hierarchy that monitors stimuli related to the depth of fluid surrounding the leech.

Interneuronal basis of the generation of related but distinct motor programs in Aplysia: implications for current neuronal models of vertebrate intralimb coordination

Coordination of two sets of movements, protraction-retraction versus opening-closing, of the feeding apparatus (the radula) in ingestive and egestive motor programs of Aplysia resembles vertebrate intralimb coordination in that the relative timing of the two sets of movements differs in the two motor programs. In both ingestion and egestion, radula protraction and retraction alternate, whereas radula closure shifts its phase relative to protraction-retraction. In egestion, the radula closes in protraction; in ingestion, the radula closes in retraction. In both ingestive and egestive motor programs elicited by the command-like neuron, cerebral-buccal interneuron-2 (CBI-2), the protraction and retraction movements are mediated by the same sets of controller interneurons. In contrast, radula closure is mediated by two controller interneurons, B20 and B40, that are preferentially active in egestion and ingestion, respectively. In egestion, B20, active in protraction, drives closure motorneuron B8 in protraction, whereas in ingestion, B40, also active in protraction, uses a functionally novel mechanism, fast inhibition and slow excitation, to drive B8 in retraction. Our findings are summarized in a neural model that permits a conceptual comparison of our model with two previous hypothetical models of intralimb coordination in spinal circuits that were proposed by Grillner (1981, 1985) and Berkowitz and Stein (1994). Although our model supports the existence of separate controllers for different movements as in the Grillner (1981, 1985) model; in terms of basic mechanisms, our model is similar to the Berkowitz and Stein (1994) model because the closure movement is mediated by separate controllers in different programs, and thus both models can be classified as recruitment models.

Interneuronal and peptidergic control of motor pattern switching in Aplysia

It has been proposed that a choice of specific behaviors can be mediated either by activation of behavior-specific higher order neurons or by distinct combinations of such neurons in different behaviors. We examined the role that two higher order neurons, CBI-2 and CBI-3, play in the selection of motor programs that correspond to ingestion and egestion, two stimulus-dependent behaviors that are generated by a single central pattern generator (CPG) of Aplysia. We found that CBI-2 could evoke either ingestive, egestive, or ambiguous motor programs depending on the regime of stimulation. When CBI-2 recruited CBI-3 firing via electrical coupling, the motor program tended to be ingestive. In the absence of CBI-3 activation, the program was usually egestive. When CBI-2 was stimulated to produce ingestive programs, hyperpolarization of CBI-3 converted the programs to egestive or ambiguous. When CBI-2 was stimulated to produce egestive or ambiguous programs, co-stimulation of CBI-3 converted them into ingestive. These findings are consistent with the idea that combinatorial commands are responsible for the choice of specific behaviors. Additional support for this view comes from the observations that appropriate stimulus conditions exist both for activation of CBI-2 together with CBI-3, and for activation of CBI-2 without a concomitant activation of CBI-3. The ability of CBI-3 to convert egestive and ambiguous programs into ingestive ones was mimicked by application of APGWamide, a neuropeptide that we have detected in CBI-3 by immunostaining. Thus combinatorial actions of higher order neurons that underlie pattern selection may involve the use of modulators released by specific higher order neurons.