© 2002 by The Society for Integrative and Comparative Biology
Decision-Making in the Leech Nervous System1
1 UCSD, Division of Biology, Neurobiology Section 0357, 9500 Gilman Dr., La Jolla, California 92093-0357
 |
SYNOPSIS |
|---|
TOP
 SYNOPSIS INTRODUCTIONCOMMAND NEURONS AND DECISION...MOTOR NEURONAL BASIS OF...THE COMBINATORIAL CODEHIERARCHICAL DECISION-MAKING...References |
|---|
Previous models of behavioral choice have described two types of hierarchy, a decision hierarchy, in which different classes of decisions are made at each level (Tinbergen, 1951
), and a behavioral hierarchy, in which one behavior will take precedence over others (Davis, 1985). Most experimental work on the neuronal basis of decision-making has focussed on the latter of these: a behavioral hierarchy is described for an animal, and the neuronal basis for this hierarchy, hypothesized to depend on inhibitory interactions, is investigated. Although the concept of "dedicated command neurons" has been useful for guiding these studies, it appears that such neurons are rare. We present evidence that in the leech, most neurons, including high-level decision neurons, are active in more than one behavior. We include data from one newly-identified neuron that elicits both swimming and crawling motor patterns. We suggest that decisions are made by a "combinatorial code": what behavior is produced depends on the specific combination of decision neurons that are active at a particular time. Finally, we discuss how decision neurons may be arranged into a decision hierarchy, with neurons at each sequential level responsible for choosing between a narrower range of behaviors. We suggest additional sensory information is incorporated at each level to inform the decision. |
INTRODUCTION |
|---|
|
TOP
SYNOPSISINTRODUCTIONCOMMAND NEURONS AND DECISION...MOTOR NEURONAL BASIS OF...THE COMBINATORIAL CODEHIERARCHICAL DECISION-MAKING...References |
|---|
Based on studies of animal behavior—mostly that of birds and fish—Nikko Tinbergen proposed that an animal chooses among behaviors in hierarchical steps (Tinbergen, 1951
). He proposed that the most general decisions (e.g., whether to migrate or to reproduce) are made in response to levels of hormones, which themselves are controlled by external signals (e.g., length of the day) and by internal states. Once a general decision is made (e.g., to reproduce), animals must decide among a number of potential behaviors (e.g., attracting a mate vs. fighting with a rival vs. building a nest) based on sensory input and previous history. These decisions, Tinbergen proposed, were likely to be made by "inhibitive interactions" between brain centers: the parts of the brain responsible for producing attract-a-mate behaviors would inhibit those responsible for fighting off a rival and for building a nest. Once one path was selected—say, building a nest—there would be other decisions like where to build it, what materials to use, how to construct the nest, etc. He saw these, too, as competing behaviors, with similar inhibitory interactions among the brain areas responsible for producing each of them. At the lowest hierarchical level—the actual production of the motor acts—Tinbergen assumed that the interactions would be cooperative, with modules of behavioral components that need to be called upon in different temporal and spatial combinations to produce each complex behavioral act. Note that in Tinbergen's model a decision at one level influences what type of sensory information will be attended to for making a decision at the next level. Thus, if a bird has decided to migrate, it will attend to sensory information about what direction to fly, but will ignore sensory cues about mating and nest building. We will return to this idea at the end of our discussion.
Another type of hierarchy has been proposed by Jack Davis and his colleagues, to explain how animals choose between different behaviors that occupy the same level in Tinbergen's hierarchy (Davis, 1985). Davis suggested that some behaviors would always override others as a result of the "inhibitive interactions" proposed by Tinbergen. One could therefore establish a behavioral hierarchy for an animal, by first finding a stimulation threshold for producing one behavior; then finding the threshold for a second, competing behavior; then presenting the two stimuli simultaneously to force a choice. Whichever behavior was produced under these conditions was considered to have the higher position in the behavioral hierarchy, and its production was thought to inhibit the other behaviors. This study laid out what would become the standard experimental procedure for identifying behavioral hierarchies.
In the initial experiments by Davis and his colleagues, the behavioral hierarchy was clear and simple: in the marine slug Pleurobranchea, egg-laying overrode feeding, which overrode righting responses, withdrawal, and mating. With further work, however, it was revealed that things were more complicated. For instance, the interactions were plastic: feeding became a much more likely choice than withdrawal as the animals got hungrier and hungrier (Davis, 1985). Recently, a model incorporating feeding and satiety into the behavioral hierarchy has been proposed (Gillette et al., 2000). Despite the complexities, however, given a motivational state, one can reliably construct a hierarchy of behavioral choices by presenting two stimuli simultaneously that would individually produce two different behaviors. Then one can begin to test for "inhibitive interactions" among the neural circuits controlling the two behaviors.
|
COMMAND NEURONS AND DECISION-MAKING |
|---|
|
TOP
SYNOPSISINTRODUCTIONCOMMAND NEURONS AND DECISION...MOTOR NEURONAL BASIS OF...THE COMBINATORIAL CODEHIERARCHICAL DECISION-MAKING...References |
|---|
A major breakthrough in the study of the neuronal basis for initiating behaviors was finding what were originally called "command neurons" based on the fact that stimulating a single one of them could produce a complex behavior (Wiersma and Ikeda, 1964
). Such neurons have been found to elicit many kinds of behaviors in a variety of animals (Kupfermann and Weiss, 1978). In many brains, particularly complex ones, multiple neurons need to be activated to produce a behavior; in such cases, these groups of neurons are called "command systems." Although demonstrating that any of these behavior-eliciting neurons are the cause of any given behavior is difficult, most investigators feel that such neurons are likely to be part of the decision-making circuitry (see Commentaries after Kupfermann and Weiss, 1978). Because there is some controversy about whether a given interneuron of this type should be called a command neuron, part of a command system, or a "command-like neuron," we have opted to use the more descriptive term "decision neuron" for the following discussion.
From studies of decision neurons in different systems, a model emerged that placed decision neurons in the middle of a neuronal hierarchy between sensory input and motor output (Fig. 1A). According to this model, decision elements collect processed sensory input and, when they are sufficiently stimulated, their activity turns on the neuronal circuit (a "pattern generator") responsible for a specific behavior. All the pattern generators use the same motor neurons. In this formulation, decision neurons are "dedicated" to a specific behavior (Krasne and Lee, 1988). That is, there are distinct decision neurons for each behavior, and each decision neuron is active during only one behavior, namely the one that is activated by stimulating that decision neuron.
|
This model led to the hypothesis that decision neurons for different behaviors inhibit each other to produce a behavioral hierarchy (Fig. 1B). Support for this hypothesis has been forthcoming. Davis found evidence for feeding inhibiting decision-makers for withdrawal (Kovac and Davis, 1980
). Since then, other instances have emerged: struggling over swimming in frog tadpoles (Green and Soffe, 1998); withdrawal or hunting behaviors over swimming in the mollusc, Clione (Panchin et al., 1994; Norekian and Satterlie, 1996); egestion over ingestion in Aplysia (Jing and Weiss, 2001); swimming over feeding in Pleurobranchea (Jing and Gillette, 2000); and feeding over tactile responses in leeches (Misell et al., 1998)
It is in this context that we began our studies of behavioral choice in the leech. We first selected two incompatible behaviors—swimming and whole-body shortening—about which we knew the neuronal circuitry to some degree (Brodfuehrer and Burns 1995; Shaw and Kristan, 1995). To understand these experiments, it is necessary to understand a bit about the leech body plan and internal anatomy.
|
MOTOR NEURONAL BASIS OF LEECH BEHAVIORS |
|---|
|
TOP
SYNOPSISINTRODUCTIONCOMMAND NEURONS AND DECISION...MOTOR NEURONAL BASIS OF...THE COMBINATORIAL CODEHIERARCHICAL DECISION-MAKING...References |
|---|
Leeches are annelid worms. They have 32 body segments, of which 4 form the head, 7 form the tail, and the remainder form 21 iterated midbody segments (Fig. 2A). To study a behavior, we first characterize the movements—the kinematics—in an intact animal. To record neural activity, we open the animal and expose part of its nervous system. Using such "semi-intact" animals, we have found motor neuron firing patterns that would produce swimming (Kristan et al., 1974
), shortening (Shaw and Kristan, 1995, 1999), and crawling (Baader, 1997) if that part of the animal were still intact. We have also found that the same firing patterns are seen in the completely isolated nervous system (Kristan and Calabrese, 1976; Eisenhart et al., 2000). In whole-body shortening, the motor neurons that excite the dorsal and ventral longitudinal muscles in each segment fire simultaneous, prolonged trains of impulses which causes co-contraction of the dorsal and ventral longitudinal muscles (Fig. 2B). In swimming, these same motor neurons fire bursts of impulses out of phase with one another, repeating at about 1 Hz (Fig. 2C). Crawling is produced by alternating bursts of longitudinal motor neurons, during the contraction phase, and circular motor neurons, during the elongation phase (Fig. 2D). In contrast to whole-body shortening, in which longitudinal motor neurons fire simultaneously in all segments, in crawling and swimming the bursts are delayed from one segment to the next, so there is a progression of bursting in longitudinal and circular motor neurons from anterior to posterior. Swimming and crawling can be elicited by activating mechanoreceptors in the posterior end of the animal; swimming is usually produced when the leech is submerged, particularly when the posterior sucker is not attached to the substrate, whereas crawling is generally produced when the leech is out of water. Shortening is readily elicited when the front end of the animal is touched.
|
Shaw and Kristan (1997)
determined the thresholds for eliciting swimming and whole-body shortening. When they then presented the stimuli simultaneously, the behavior that always predominated was shortening. In fact, if a swimming leech was stimulated at the same stimulus intensity that produced shortening in a resting leech, the swimming stopped within a cycle and the animal shortened (Shaw and Kristan, 1997). A summary of the known neuronal circuits for these two behaviors is shown in Figure 3A, B. Whole-body shortening is produced by two parallel networks, a weak but fast one that gets the shortening started, and a relatively slower but more prolonged and more effective one that produces most of the behavioral response. Both are activated by mechanosensory neurons in the front of the animal that respond to light touch (T cells), pressure (P cells), and noxious stimuli (N cells). The known swimming network is a five-layered, mostly feed-forward network, starting with T and P mechanosensors in the back end of the animals. Because the Trigger and Gating interneurons for swimming appear to be "decision neurons," those were the focus of our attention in investigating how the shortening circuitry affected ongoing swimming. To do these experiments we used a semi-intact preparation.
|
The first neuron we chose to study was cell 204, because it has a strong ability to initiate and maintain the swimming motor pattern (Weeks and Kristan, 1978
). As shown in Figure 4A, cell 204 is strongly inhibited by tactile stimuli that terminate swimming and turn on shortening. This is exactly what the decision model predicted: turning on one behavior (shortening) inhibited the decision neurons for a second behavior (swimming).
|
In contrast to cell 204, many other swim neurons were excited during whole-body shortening. These included some of the "swim CPG" interneurons, which was not surprising because Getting and Dekin showed decades ago (Getting and Dekin, 1985
) that CPG interneurons that were active in an oscillatory mode when a Tritonia swims are also active tonically when the animal withdraws from a tactile stimulus. What was surprising, however, was that a second "swim gating" interneuron was excited during whole-body shortening (Fig. 4B) and that, as summarized in Figure 4C, the "swim trigger" interneurons were also excited during shortening. In other words, many of the neurons involved in swimming, even decision neurons are multi-functional, i.e., they are active in more than one behavior. In our recent work, we have identified yet another "swim initiator" that is multi-functional. This previously-undescribed neuron, which we call R3b1, has its soma in the subesophageal ganglion of the head, and its axon projects down the nerve cord to the tail. Stimulation of the neuron with intracellular current injection in an isolated nerve cord sometimes elicited the swimming motor pattern (Fig. 5A). Often, however, the same stimulation elicited a different motor pattern, one that resembles that of crawling (Fig. 5B). The two behaviors can be elicited by subsequent stimulation in the same isolated nerve cord, with no experimentally-induced changes to the preparation other than repeated stimulation of the one neuron. This result is even more surprising than the previous findings, because not only is this new decision neuron active during more than one behavior, but also it by itself can initiate two distinct behaviors. We are currently examining factors that might contribute to this decision, but the fact that two different behaviors are elicited by the same stimulation in an isolated nervous system indicates that some internal state variables must play a role.
|
|
THE COMBINATORIAL CODE |
|---|
|
TOP
SYNOPSISINTRODUCTIONCOMMAND NEURONS AND DECISION...MOTOR NEURONAL BASIS OF...THE COMBINATORIAL CODEHIERARCHICAL DECISION-MAKING...References |
|---|
If so many neurons are active during different behaviors, how does a leech decide what to do in any given situation? Figure 6 indicates that these decisions may be made by a combinatorial code: whether swimming, crawling, shortening or bending occurs depends on the combination of decision-making neurons active. We know, for instance, that cell 204 is active during swimming and inhibited during shortening. Other experiments have shown that this neuron is not active during local bending (Lockery and Kristan, 1990
), but is active during crawling (Kristan et al., 1988). In addition, cells 61, SE1, and Tr1 are active during both swimming and shortening (Shaw and Kristan, 1997), but are not active during crawling or local bending. Data in Figure 5 shows that cell R3b1 activates both swimming and crawling. Further experiments (data not shown) indicate that this neuron is activated when these behaviors are turned on by other means (such as pinching the animal in the rear end), and is shut off during shortening. We assume that other such decision interneurons, not yet discovered, will be active during other combinations of behaviors. Hence, no single interneuron is a good predictor of when a behavior will occur. Instead, one needs to examine the activity of the whole population of interneurons.
|
|
HIERARCHICAL DECISION-MAKING REVISITED |
|---|
|
TOP
SYNOPSISINTRODUCTIONCOMMAND NEURONS AND DECISION...MOTOR NEURONAL BASIS OF...THE COMBINATORIAL CODEHIERARCHICAL DECISION-MAKING...References |
|---|
Deciding to do something is often a complicated process that takes into account many different variables, such as motivational state and recent experience. Not only do these variables alter the threshold for behavioral initiation, but they can also influence what sensory information the animal acquires. For example, hungry leeches are more likely to move around spontaneously and will respond to lower threshold stimuli than leeches that have recently eaten (Willard, 1981
). Furthermore, it seems hungry leeches actively position themselves at the water surface so that they will be more likely to receive stimulation when potential prey enters the water (Dickinson and Lent, 1984). This is a good example of Tinbergen's decision-making hierarchy, in which sensory information is utilized at different levels. The leech first decides that it will try to find a food source, then positions itself to attend to relevant stimuli before making the next decision about which direction to swim. As described above, previous models of the neural architecture underlying behavioral choice have usually assumed a serial progression from sensory input to decision neurons to motor output, with sensory processing occurring before a decision is made. The possibility of sensory input to neurons downstream of the decision neurons has usually been ignored, or thought only to modulate the chosen behavior. But in Tinbergen's model, sensory information is used at several points in hierarchical decision-making: after a bird decides to reproduce, additional sensory information determines whether it builds a nest or attracts a mate.
We have recently acquired additional, preliminary, data for the type of hierarchical decision-making proposed by Tinbergen in the nervous system of the leech. These data suggest that sensory information may feed in downstream of cell R3b1 to help the leech decide whether to swim or to crawl. It appears that R3b1 commands the leech to do one of these locomotory behaviors, but the decision between the two behaviors is made downstream.
A summary of our data from different leech interneurons and hypotheses about their roles in hierarchical decision making is shown in Table 1. At the top level of the hierarchy, the results of Shaw and Kristan (1997) suggest that previously described swim-trigger neurons, which are also active during shortening, may command the leech, "Do something!" At the next step, R3b1 commands the leech, "Swim or crawl!" Next, cell 204 says, "Swim!" and activates the CPG elements that produce this behavior (Weeks and Kristan, 1978). Some of these CPG elements command behavioral modules, such as "dorsal contraction in segment 3" and "circular contraction in segment 17," which may also be activated in other behaviors.
|
We are currently investigating what types of sensory input may influence the leech's decision to swim or crawl. One likely candidate is serotonin, which is elevated in hungry leeches and in leeches that swim frequently (Willard, 1981
). Interestingly, whereas frequent swimmers have the highest levels of serotonin, leeches with intermediate levels are more likely to crawl than those with low levels. Sensory information about water level is also likely to play a role in the decision to swim or crawl, since leeches cannot swim in shallow water. Another possible input is from thermoreceptors, since applying heat is an effective stimulus for crawling (Eisenhart et al., 2000).
It makes good evolutionary sense for an animal to incorporate diverse sensory information at multiple decision points before making a final behavioral choice. Unfortunately, this complicates the task of researchers, who must record the activity of many neurons over time to understand the decision process. We hope that the recent development of voltage-sensitive dyes for recording from populations of neurons (Cacciatore et al., 1999) will help us in this endeavor.
|
ACKNOWLEDGMENTS |
|---|
This work was supported by NIH Grants MH43396 and NS35336 (WBK); and NRSA MH12029 and NIH Training Grant NS07220 (TE).
|
FOOTNOTES |
|---|
1 From the Symposium Recent Advances in Neurobiology presented at the Annual Meeting of the Society for Integrative and Comparative Biology, 2–6 January 2002, at Anaheim, California.
2 E-mail: wkristan{at}ucsd.edu
|
References |
|---|
|
TOP
SYNOPSISINTRODUCTIONCOMMAND NEURONS AND DECISION...MOTOR NEURONAL BASIS OF...THE COMBINATORIAL CODEHIERARCHICAL DECISION-MAKING...References |
|---|
Baader, A. P. 1997. Interneuronal and motor patterns during crawling behavior of semi-intact leeches. J. Exp. Biol, 200:1369-1381.[Abstract]
Brodfuehrer, P. D., and A. Burns. 1995. Neuronal factors influencing the decision to swim in the medicinal leech. Neurobiol. Learn. Mem, 63:192-199.[CrossRef][Web of Science][Medline]
Brodfuehrer, P. D., E. A. Debski, B. A. O'Gara, and W. O. Friesen. 1995. Neuronal control of leech swimming. J. Neurobiol, 27:403-418.[CrossRef][Web of Science][Medline]
Cacciatore, T. W., P. D. Brodfuehrer, J. E. Gonzalez, T. Jiang, S. R. Adams, R. Y. Tsien, W. B. Kristan Jr., and D. Kleinfeld. 1999. Identification of neural circuits by imaging coherent electrical activity with FRET-based dyes. Neuron, 23:449-459.[CrossRef][Web of Science][Medline]
Davis, W. J. 1985. Neural mechanisms of behavioral plasticity in an invertebrate model system. In A. Selverston (ed.), Model neural networks and behavior, pp. 263–282. Plenum, New York.
Dickinson, M. H., and C. M. Lent. 1984. Feeding behavior in the leech, Hirudo medicinalis L. J. Comp. Physiol. A, 154:449-455.[CrossRef]
Eisenhart, F. J., T. W. Cacciatore, and W. B. Kristan Jr. 2000. A central pattern generator underlies crawling in the medicinal leech. J. Comp. Physiol. A, 186:631-643.[CrossRef][Medline]
Getting, P. A., and M. S. Dekin. 1985. Tritonia swimming: A model system for integration without rhythmic motor systems. In A. Selverston (ed.), Model neural networks and behavior, pp. 3–20. Plenum, New York.
Gillette, R., R. C. Huang, N. Hatcher, and L. L. Moroz. 2000. Cost-benefit analysis potential in feeding behavior of a predatory snail by integration of hunger, taste, and pain. Proc. Natl. Acad. Sci. U.S.A, 97:3585-3590.
Green, C. S., and S. R. Soffe. 1998. Roles of ascending inhibition during two rhythmic motor patterns in Xenopus tadpoles. J. Neurophysiol, 79:2316-2328.
Jing, J., and R. Gillette. 2000. Escape swim network interneurons have diverse roles in behavioral switching and putative arousal in Pleurobranchaea. J. Neurophysiol, 83:1346-1355.
Jing, J., and K. R. Weiss. 2001. Neural mechanisms of motor program switching in Aplysia. J. Neurosci, 21:7349-7362.
Kovac, M. P., and W. J. Davis. 1980. Neural mechanism underlying behavioral choice in Pleurobranchaea. J. Neurophysiol, 43:469-487.
Krasne, F. B., and S. C. Lee. 1988. Response-dedicated trigger neurons as control points for behavioral actions: Selective inhibition of lateral giant command neurons during feeding in crayfish. J. Neurosci, 8:3703-3712.[Abstract]
Kristan, W. B., G. S. Stent, and C. A. Ort. 1974. Neuronal control of swimming in the medicinal leech. I. Dynamics of the swimming rhythm. J. Comp. Physiol, 94:97-119.
Kristan, W. B., Jr., and R. L. Calabrese. 1976. Rhythmic swimming activity in neurones of the isolated nerve cord of the leech. J. Exp. Biol, 65:643-668.
Kristan, W. B., Jr., G. Wittenberg, M. P. Nusbaum, and W. Stern-Tomlinson. 1988. Multifunctional interneurons in behavioral circuits of the medicinal leech. Experientia, 44:383-389.[CrossRef][Web of Science][Medline]
Kupfermann, L., and K. R. Weiss. 1978. The command neuron concept. Behav. Brain Sci, 1:3-39.
Lockery, S. R., and W. B. Kristan Jr. 1990. Distributed processing of sensory information in the leech. II. Identification of interneurons contributing to the local bending reflex. J. Neurosci, 10:1816-1829.[Abstract]
Misell, L. M., B. K. Shaw, and W. B. Kristan Jr. 1998. Behavioral hierarchy in the medicinal leech, Hirudo medicinalis: Feeding as a dominant behavior. Behav. Brain Res, 90:13-21.[CrossRef][Web of Science][Medline]
Norekian, T. P., and R. A. Satterlie. 1996. Whole body withdrawal circuit and its involvement in the behavioral hierarchy of the mollusk Clione limacina. J Neurophysiol, 75:529-537.
Panchin, Y. V., G. N. Gamkrelidze, L. B. Popova, T. G. Deliagina, G. N. Orlovsky, and Y. I. Arshavsky. 1994. Neuronal basis of hunting and feeding behavior in the pteropod mollusc Clione limacina. Nether. J. Zool, 44:170-183.
Shaw, B. K., and W. B. Kristan Jr. 1995. The whole-body shortening reflex of the medicinal leech: Motor pattern, sensory basis, and interneuronal pathways. J. Comp. Physiol. A, 177:667-681.[Medline]
Shaw, B. K., and W. B. Kristan Jr. 1997. The neuronal basis of the behavioral choice between swimming and shortening in the leech: Control is not selectively exercised at higher circuit levels. J. Neurosci, 17:786-795.
Shaw, B. K., and W. B. Kristan, Jr. 1999. Relative roles of the S cell network and parallel interneuronal pathways in the whole-body shortening reflex of the medicinal leech. J. Neurophysiol, 82:1114-1123.
Tinbergen, N. 1951. The study of instinct. Clarenden Press, Oxford.
Weeks, J. C., and W. B. Kristan Jr. 1978. Initiation, maintenance, and modulation of swimming in the medicinal leech by the activity of a single neurone. J. Exp. Biol, 77:71-88.
Wiersma, C. A. G., and K. Ikeda. 1964. Interneurons commanding swimmeret movements in the crayfish, Procambarus clarkii (Girard). Comp. Biochem. Physiol, 12:509-525.[Medline]
Willard, A. L. 1981. Effects of serotonin on the generation of the motor program for swimming by the medicinal leech. J. Neurosci, 1:936-944.[Web of Science][Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  W. H. Liden and J. Herberholz Behavioral and neural responses of juvenile crayfish to moving shadows J. Exp. Biol., May 1, 2008; 211(9): 1355 - 1361. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 T. J. Prescott Forced Moves or Good Tricks in Design Space? Landmarks in the Evolution of Neural Mechanisms for Action Selection Adaptive Behavior, March 1, 2007; 15(1): 9 - 31. [Abstract] [PDF]
|
|
|
|
 |
|
 K. L. Briggman, H. D. I. Abarbanel, and W. B. Kristan Jr. Optical Imaging of Neuronal Populations During Decision-Making Science, February 11, 2005; 307(5711): 896 - 901. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||