- Email: [email protected]
A paradigm for position learning in the crayfish claw
Brain Research, 134 (1977) 185 190 c() Elsevier/North-Holland Biomedical Press
185
A paradigm for position learning in the crayfish claw
CARL E. STAFSTROM and GEORGE L. GERSTEIN Departments of Biophysics and of Physiology, University o] Pennsylvania School of Medicine, Philadelphia, Pa. 19104 (U.S.A.)
(Accepted June 9th, 1977)
Several invertebrate preparations capable of modifiable behavior, especially avoidance conditioning, have been reported. In mollusks, for instance, habituation of the gill withdrawal reflex lo, and rapid food-avoidance learning using such aversive stimuli as electrical shocklZ, 13 or CO2 poisoning 4, have been shown. In insects, Horridge 6,v has shown that cockroaches can be trained to maintain a leg position above a specified level, below which the animal would receive an aversive shock. Eisenstein and Cohen 3 have presented evidence that the neurons involved in this leg lift learning are located in the insect's prothoracic ganglion. These workers hypothesized that the acquisition of specific leg position by a yoked control animal (which received non-position contingent shocks) depended upon the number of specific leg position/shock associations it had experienced. Results of Disterhoft et al. 2 support this hypothesis. Hoyle 9 demonstrated that the excitor motoneuron innervating the anterior coxal adductor muscle responsible for leg raising in locusts could be trained to maintain a criterion firing level, using a pattern of contingent reinforcement similar to that of Horridge. Evidence for learning the reverse paradigm (i.e. insect required to lower leg to avoid shock) is somewhat equivocablO 4, but it has recently been shown that locust anterior coxal adductor motoneurons can be trained to either increase or decrease their firing rates z3. The physiological mechanisms underlying these modifiable behaviors are not fully understood, but an increase in motoneuron membrane resistance and a decrease in its potassium conductance seem to be associated with uplearning, while a decrease in membrane resistance and an increase in potassium conductance have been implicated in down-learning 2z. We have examined the possibility that another invertebrate could be similarly treated in order to create a preparation useful for the neurophysiological study of learning. The crayfish claw is comprised of a dactylopodite which moves relative to a propodite in a single plane. Movement and reflex behavior of the claw, which have been extensively studiedm~,l.~,~7-19, are controlled by two muscles (an opener and a closer) and 5 motoneurons (fast and slow closer excitors, closer inhibitor, opener excitor and opener inhibitor). The relative ease of accessibility to both the neural and muscular components of the system, coupled with the demonstration that the claw possesses plastic properties 6, indicates that such a preparation could prove valuable in
186
CONTROLCLAW~/
I
tl
Pos,T,oNALcw DISABLE/ENABLE
1,
t
I DIAMPLI FFERENTI POSITION FIERAIL t THRESHOWJ 1
TAPERECORDER i Fig. 1. Schematic plan of the experimental set-up. Photoelectric position sensors on each dactylopodite provide positional information to a chart recorder and, in the case of the positional claw, also to circuitry which sends shock trains to both claws as long as the positional daetylopodite exceeds its threshold 0p. The disable/enable switch allows bypass of the shock delivery circuitry. P, propodite; D, dactylopodite.
revealing the neurophysiological mechanisms underlying learning. The paradigm reported here provides a promising situation in which most of the relevant neuronal activity might be monitored during learning. In our preparation, a crayfish was restrained, dorsal side down, so that only the dactylopodites were allowed to move, this movement being monitored by photoelectric position sensors (Fig. 1). Two position thresholds were defined, 0o for the positional claw and 0e for the control claw. These thresholds were chosen to be some fraction of the fully-open position of each claw. Typical values were 20-25 ~, where 100 ~ represented the fully-open position, Bipolar chlorided silver wire electrodes (250 #m, Insulex insulation) were inserted approximately 1 mm through small holes drilled in the propodite lateral margin over the closer muscle. Electronic circuitry provided a train of shocks (1-5 V, 10/sec) to the electrodes in both claws whenever the positional claw's dactylopodite moved above its designated threshold, 00. Thus, the control claw received shocks whenever the positional claw did, but irrespective of its own position. In this way, each animal acted simultaneously as experiment and control. There are several advantages to this approach. Besides reducing the spatial complications of using another animal as a yoked control, problems of hormonal and systemic
187 100
PositionalClaw
PositionalClaw
75
50
25
tO. 0 100
~...
~,f f / f z / ~ j ~
y°Open
~
1%Open
ControlClaw
ControlClaw
75 50 : :3:: ,..
25 :)i?:'?'.?;!i:-??i": 30
%e" Open = 60
90
Time (Min)
120
~
1 30
% 60
Open 90
120
Time (Min)
Fig. 2. Typical data from two crayfish (A, left and B, right). Mean positional (top) and control (bottom) dactylopodite position for each 5 min interval is plotted. 0 ~ represents full closure, 100~ represents the fully open position. Threshold levels (0) are indicated ; 0p was determined before the experiment, 0e was later constructed to be the same proportion of opening as 0v. Variances in position are insignificant unless plotted. Control period (shock disabled), coarse stippling; training period (shock enabled), fine stippling; test period (shock disabled), diagonal lines. effects are eliminated. Horridge 9 recognized the utility of such a single animal preparation. Experiments were divided into three periods: (1) During the control period of 30-35 min, no shocks were administered to the claws. (2) During the 25-30 rain training period, shocks were delivered to both claws whenever the positional dactylopodite opened past 0p. (3) A test period of approximately 60 min followed, during which neither claw received shocks. Durations of the periods were later modified to 20 min each. Positions of both dactylopodites were written out on a chart recorder. An additional trace indicated when the positional dactylopodite exceeded 0p. Crossings of 0e were not determined electronically, but were later constructed from the chart recorder printout. Fig. 2 shows data histograms from two typical experiments (A and B). Mean dactylopodite position was measured and plotted using 5 rain epochs in order to make possible statistical comparisons between position of the two claws. In each experiment, both positional and control dactylopodites remained above the selected threshold levels throughout the control periods in both experiments. Under the conditions of the control period no fixed relationship was observed between movements of the two
188 claws. Sometimes movements were in unison, sometimes not. Reflex movement t~) touch stimulation of hair receptors on the claw could be elicited from one cla~ at ~ time; touch stimulation of the abdomen sometimes elicited movements of both claws. When the shock was enabled at the beginning of the training period, the histograms show that the animal rapidly developed a behavior to keep the shock off, In both experiments shown, the positional dactylopodite quickly fell and remained below 0)), whereas the control dactylopodite did not attain or hold any particular level. This pattern of behavior persisted through the test periods, when the shock was disabled. Thus, for the positional claw, mean dactylopodite position during the training and test periods differed significantly from the mean dactylopodite position during lhe control periods (unpaired t-test, 0.01 level). For the control claw, the mean positions did not differ significantly (unpaired t-test, 0.05 level). Comparison of Figs. 2A and 2B will illustrate variation typical of the experiments reported here. After about 100 min, the positional dactylopodite in Fig. 2A exceeds 0p, receives no shocks, and maintains its position above threshold. In Fig. 2B, the control dactylopodite exhibits substantially different positions in each 5 rain epoch during the training period, but during the test period it assumes a stable position very TABLE l Summary o f results Total experiments
Learned in one session
Added with two sessions
Faihtre
19
II
5
3
6
4
--
2
3
-
3
I
-
I
0
....
3
(A) I. Standard paradigm Procambarus elarkii and cambarus diogenes
Close-learning Shock to closer muscle of claw 30 min control/25 min train/60 min test (A) IL Revised paradigm Proeambarus elarkii
Close-learning Shock to closer muscle of claw 20 min control/20 min train/20 min test (B) Alternate paradigm Cambarus diogenes
Close-learning a. Tail shocked 6 b, First pair walking legs shocked 2 30 min control/25 min train/60 min test (C) Open-learning experiments Proeambarus elarkii and Cambarus diogenes
Shock to closer muscle of claw
3
189 similar to that of the control period. This dactylopodite briefly attains a resting position below 0e halfway in the test period. Whether such variations are due to exploration, fatigue, or some other process, or whether they merely represent individual variation, cannot be determined from these experiments. Further investigation is needed to determine how long the animal can retain such position learning and how it behaves upon exchange of positional and control claws in subsequent training sessions. Additional experiments have confirmed results presented in a preliminary report of this work 17. Of 25 Procambarus clarkii and Cambarus diogenes specimens tested, 15 succeeded at the 'close-learning' task after one training session (Table i). There was no evidence that either the right or left claw learned the paradigm better. Five additional animals learned the paradigm after two consecutive training sessions, separated by a short rest period (20 rain). Five crayfish failed to learn the position at all. Due to the electrode arrangement described above, the precise effects of the 'aversive' stimulus could not be delineated. Some sensory components must certainly be involved, as well as effects of the shock on the muscle itself, but the contributions of each are not known. To control for the specific effects of the shock on claw musculature, we tried some experiments in which other anatomical areas of the crayfish were shocked instead. Preliminary results indicate that even when the tail or walking legs are shocked, the claws can exhibit position learning, although the success/failure ratio is much lower than when the claws themselves are stimulated (Table 1). In additional experiments, the motor axons alone were severed, carefully avoiding the possibility of 'welding' sensory and motor axons. Under these conditions, no tonic spontaneous neural activity can reach the muscle; the claw is flaccid. Muscle contraction induced by a train of shocks relaxes a few seconds after end of stimulation, with no net change of position. Attempts to reverse the paradigm, i.e. the claw is shocked if it closes beyond the designated threshold ('open-learning'), have not been successful so far (three failures, no successes). It may be possible, as in the locust 2'~, to use a paradigm in which the crayfish must learn to maintain its dactylopodite within a particular range. Although this report does not address the mechanisms underlying the change of claw position, we must in the future consider both central and peripheral contributions. A particularly relevant peripheral mechanism is the catch property that has been documented by Wilson and coworkers2O, 2~. They showed that a short, high frequency burst of stimulation could considerably enhance the tension produced by a continuing slow stimulation. In our conditions, the contraction of the positional claw closer muscle produced by an unchanging tonic motor nerve input could be enhanced for some minutes by the electrical stimulus. This known mechanism, however, does not explain the long duration of the effect during the test period nor the lack of position change exhibited by the control claw; thus, some central changes seem highly likely. The prospect of recording from many of the relevant neurons while an animal is learning remains an exciting one. This preparation offers a promising model system for studying the cellular basis of learning in an intact, behaving animal.
190 This work was supported by U S P H S G r a n t NS 05606. The a u t h o r s would like to t h a n k Dr. Bruce Lindsey for suggestions and assistance.
1 Atwood, H. L. and Wiersma, C. A. G., Command interneurons in the crayfish central nervous system, J. exp. BioL, 46 (1967) 249-261. 2 Disterhoft, J. F., Nurnberger, J. and Coming, W. C., 'P-R' differences in intact cockroaches as a function of testing interval, Psychon. Sci., 12 (1968) 205-206. 3 Eisenstein, E. M. and Cohen, M. J., Learning in an isolated prothoracic insect ganglion, Animal Behav., 13 (1965) 104-108. 4 Gelperin, A., Rapid food aversion learning by a terrestrial mollusk, Science, 189 (1975) 567--570. 5 Gerstein, G. L., Schuster, R. H. and Wiens, T. J., Neural activity and behavioral plasticity in the crayfish claw. In J. A. Kiger (Ed.), The Biology of Behavior, Oregon State Univ. Press, Corvallis, 1972, pp. 67-82. 6 Horridge, G. A., Learning of leg position by the ventral nerve cord in headless insects. Proc. Roy. Soc. B, 157 (1962) 33-52. 7 Horridge, G. A., Learning of leg position by headless insects, Nature (Lond.), 193 (1962) 697-698. 8 Horridge, G. A., The electrophysiological approach to learning in isolatable ganglia. In W. H. Thorpe and D. Davenport (Eds.), Animal Behaviour, SuppL 1, 1965. 9 Hoyle, G., Neurophysiologieal studies of learning in headless insects. In J. E. Treherne and J. W. L. Beament (Eds.), The Physiology of the lnsect Central Nervous System, Academic Press, London, 1965. 10 Kandel, E. R., An invertebrate system for the cellular analysis of simple behaviors. In F. O. Schmitt and F. G. Worden (Eds.), The Neurosciences: Third Study Program, MIT Press, Cambridge, 1974, pp. 347-370. 11 Lindsey, B. G. and Gerstein, G. L., Spatio-temporal characterization of crayfish claw efferents' firing patterns during step movements of the dactylopodite, Abstr., Soc. Neurosci., 6 (1976) 330. 12 Mpitsos, G. J. and Collins, S. D., Learning: rapid aversive conditioning in the gastropod mollusk Pleurobranchaea, Science, 188 (1975) 954-957. 13 Mpitsos, G. J. and Davis, W. J., Learning: classical and avoidance conditioning in the mollusk Pleurobranchaea, Science, 180 (1973) 317-320. 14 Pritchatt, D., Avoidance of electric shock by the cockroach Periplaneta americana, Anita Behav. 16 (1968) 178-185. 15 Smith, D. O., Central nervous control of excitatory and inhibitory neurons of the opener muscle of the crayfish claw, J. NeurophysioL, 37 (1974) 108-118. 16 Stafstrom, C. E. and Gerstein, G. L., A paradigm for position learining in the crayfish claw, Abstr., Soc. Neurosci., 2 (1976) 437. 17 Wiens, T. J. and Atwood, H. L., Intracellular studies of crayfish claw motoneurons, Abstr., Soc. Neurosci., 2 (1976) 260. 18 Wiens, T. J. and Gerstein G. L., Cross connections among crayfish claw efferents, .l. NeurophysioL, 38 (1975) 909-921. 19 Wiens, T. J. and Gerstein, G. L., Reflex pathways in the crayfish claw, J. ~mp. PhvsioL, At07 (1976) 309-326. 20 Wilson, D. M. and Larimer, J. L., The catch property of ordinary muscle, Proc. nat. Acad. Sci. (Wash.), 61 (1968) 909-916. 21 Wilson, D. M., Smith, D. O. and Dempster, P., Length and tension hysteresis during sinusoidal and step function stimulation of arthropod muscle, Amer. J. PhysioL, 218 (1970) 916-922: 22 Woollacott, M. H. and Hoyle, G., Membrane resistance changes associated with single, identified neuron learning, Abstr., Soc. NeuroscL 6 (1976) 339. 23 Woollacott, M. H., Tosney, T. and Hoyle, G., Pacemaker changes as a principal basis tbr insect leg learning, Abstr. Soc. Neurosci., 1 (1975) 501.
185
A paradigm for position learning in the crayfish claw
CARL E. STAFSTROM and GEORGE L. GERSTEIN Departments of Biophysics and of Physiology, University o] Pennsylvania School of Medicine, Philadelphia, Pa. 19104 (U.S.A.)
(Accepted June 9th, 1977)
Several invertebrate preparations capable of modifiable behavior, especially avoidance conditioning, have been reported. In mollusks, for instance, habituation of the gill withdrawal reflex lo, and rapid food-avoidance learning using such aversive stimuli as electrical shocklZ, 13 or CO2 poisoning 4, have been shown. In insects, Horridge 6,v has shown that cockroaches can be trained to maintain a leg position above a specified level, below which the animal would receive an aversive shock. Eisenstein and Cohen 3 have presented evidence that the neurons involved in this leg lift learning are located in the insect's prothoracic ganglion. These workers hypothesized that the acquisition of specific leg position by a yoked control animal (which received non-position contingent shocks) depended upon the number of specific leg position/shock associations it had experienced. Results of Disterhoft et al. 2 support this hypothesis. Hoyle 9 demonstrated that the excitor motoneuron innervating the anterior coxal adductor muscle responsible for leg raising in locusts could be trained to maintain a criterion firing level, using a pattern of contingent reinforcement similar to that of Horridge. Evidence for learning the reverse paradigm (i.e. insect required to lower leg to avoid shock) is somewhat equivocablO 4, but it has recently been shown that locust anterior coxal adductor motoneurons can be trained to either increase or decrease their firing rates z3. The physiological mechanisms underlying these modifiable behaviors are not fully understood, but an increase in motoneuron membrane resistance and a decrease in its potassium conductance seem to be associated with uplearning, while a decrease in membrane resistance and an increase in potassium conductance have been implicated in down-learning 2z. We have examined the possibility that another invertebrate could be similarly treated in order to create a preparation useful for the neurophysiological study of learning. The crayfish claw is comprised of a dactylopodite which moves relative to a propodite in a single plane. Movement and reflex behavior of the claw, which have been extensively studiedm~,l.~,~7-19, are controlled by two muscles (an opener and a closer) and 5 motoneurons (fast and slow closer excitors, closer inhibitor, opener excitor and opener inhibitor). The relative ease of accessibility to both the neural and muscular components of the system, coupled with the demonstration that the claw possesses plastic properties 6, indicates that such a preparation could prove valuable in
186
CONTROLCLAW~/
I
tl
Pos,T,oNALcw DISABLE/ENABLE
1,
t
I DIAMPLI FFERENTI POSITION FIERAIL t THRESHOWJ 1
TAPERECORDER i Fig. 1. Schematic plan of the experimental set-up. Photoelectric position sensors on each dactylopodite provide positional information to a chart recorder and, in the case of the positional claw, also to circuitry which sends shock trains to both claws as long as the positional daetylopodite exceeds its threshold 0p. The disable/enable switch allows bypass of the shock delivery circuitry. P, propodite; D, dactylopodite.
revealing the neurophysiological mechanisms underlying learning. The paradigm reported here provides a promising situation in which most of the relevant neuronal activity might be monitored during learning. In our preparation, a crayfish was restrained, dorsal side down, so that only the dactylopodites were allowed to move, this movement being monitored by photoelectric position sensors (Fig. 1). Two position thresholds were defined, 0o for the positional claw and 0e for the control claw. These thresholds were chosen to be some fraction of the fully-open position of each claw. Typical values were 20-25 ~, where 100 ~ represented the fully-open position, Bipolar chlorided silver wire electrodes (250 #m, Insulex insulation) were inserted approximately 1 mm through small holes drilled in the propodite lateral margin over the closer muscle. Electronic circuitry provided a train of shocks (1-5 V, 10/sec) to the electrodes in both claws whenever the positional claw's dactylopodite moved above its designated threshold, 00. Thus, the control claw received shocks whenever the positional claw did, but irrespective of its own position. In this way, each animal acted simultaneously as experiment and control. There are several advantages to this approach. Besides reducing the spatial complications of using another animal as a yoked control, problems of hormonal and systemic
187 100
PositionalClaw
PositionalClaw
75
50
25
tO. 0 100
~...
~,f f / f z / ~ j ~
y°Open
~
1%Open
ControlClaw
ControlClaw
75 50 : :3:: ,..
25 :)i?:'?'.?;!i:-??i": 30
%e" Open = 60
90
Time (Min)
120
~
1 30
% 60
Open 90
120
Time (Min)
Fig. 2. Typical data from two crayfish (A, left and B, right). Mean positional (top) and control (bottom) dactylopodite position for each 5 min interval is plotted. 0 ~ represents full closure, 100~ represents the fully open position. Threshold levels (0) are indicated ; 0p was determined before the experiment, 0e was later constructed to be the same proportion of opening as 0v. Variances in position are insignificant unless plotted. Control period (shock disabled), coarse stippling; training period (shock enabled), fine stippling; test period (shock disabled), diagonal lines. effects are eliminated. Horridge 9 recognized the utility of such a single animal preparation. Experiments were divided into three periods: (1) During the control period of 30-35 min, no shocks were administered to the claws. (2) During the 25-30 rain training period, shocks were delivered to both claws whenever the positional dactylopodite opened past 0p. (3) A test period of approximately 60 min followed, during which neither claw received shocks. Durations of the periods were later modified to 20 min each. Positions of both dactylopodites were written out on a chart recorder. An additional trace indicated when the positional dactylopodite exceeded 0p. Crossings of 0e were not determined electronically, but were later constructed from the chart recorder printout. Fig. 2 shows data histograms from two typical experiments (A and B). Mean dactylopodite position was measured and plotted using 5 rain epochs in order to make possible statistical comparisons between position of the two claws. In each experiment, both positional and control dactylopodites remained above the selected threshold levels throughout the control periods in both experiments. Under the conditions of the control period no fixed relationship was observed between movements of the two
188 claws. Sometimes movements were in unison, sometimes not. Reflex movement t~) touch stimulation of hair receptors on the claw could be elicited from one cla~ at ~ time; touch stimulation of the abdomen sometimes elicited movements of both claws. When the shock was enabled at the beginning of the training period, the histograms show that the animal rapidly developed a behavior to keep the shock off, In both experiments shown, the positional dactylopodite quickly fell and remained below 0)), whereas the control dactylopodite did not attain or hold any particular level. This pattern of behavior persisted through the test periods, when the shock was disabled. Thus, for the positional claw, mean dactylopodite position during the training and test periods differed significantly from the mean dactylopodite position during lhe control periods (unpaired t-test, 0.01 level). For the control claw, the mean positions did not differ significantly (unpaired t-test, 0.05 level). Comparison of Figs. 2A and 2B will illustrate variation typical of the experiments reported here. After about 100 min, the positional dactylopodite in Fig. 2A exceeds 0p, receives no shocks, and maintains its position above threshold. In Fig. 2B, the control dactylopodite exhibits substantially different positions in each 5 rain epoch during the training period, but during the test period it assumes a stable position very TABLE l Summary o f results Total experiments
Learned in one session
Added with two sessions
Faihtre
19
II
5
3
6
4
--
2
3
-
3
I
-
I
0
....
3
(A) I. Standard paradigm Procambarus elarkii and cambarus diogenes
Close-learning Shock to closer muscle of claw 30 min control/25 min train/60 min test (A) IL Revised paradigm Proeambarus elarkii
Close-learning Shock to closer muscle of claw 20 min control/20 min train/20 min test (B) Alternate paradigm Cambarus diogenes
Close-learning a. Tail shocked 6 b, First pair walking legs shocked 2 30 min control/25 min train/60 min test (C) Open-learning experiments Proeambarus elarkii and Cambarus diogenes
Shock to closer muscle of claw
3
189 similar to that of the control period. This dactylopodite briefly attains a resting position below 0e halfway in the test period. Whether such variations are due to exploration, fatigue, or some other process, or whether they merely represent individual variation, cannot be determined from these experiments. Further investigation is needed to determine how long the animal can retain such position learning and how it behaves upon exchange of positional and control claws in subsequent training sessions. Additional experiments have confirmed results presented in a preliminary report of this work 17. Of 25 Procambarus clarkii and Cambarus diogenes specimens tested, 15 succeeded at the 'close-learning' task after one training session (Table i). There was no evidence that either the right or left claw learned the paradigm better. Five additional animals learned the paradigm after two consecutive training sessions, separated by a short rest period (20 rain). Five crayfish failed to learn the position at all. Due to the electrode arrangement described above, the precise effects of the 'aversive' stimulus could not be delineated. Some sensory components must certainly be involved, as well as effects of the shock on the muscle itself, but the contributions of each are not known. To control for the specific effects of the shock on claw musculature, we tried some experiments in which other anatomical areas of the crayfish were shocked instead. Preliminary results indicate that even when the tail or walking legs are shocked, the claws can exhibit position learning, although the success/failure ratio is much lower than when the claws themselves are stimulated (Table 1). In additional experiments, the motor axons alone were severed, carefully avoiding the possibility of 'welding' sensory and motor axons. Under these conditions, no tonic spontaneous neural activity can reach the muscle; the claw is flaccid. Muscle contraction induced by a train of shocks relaxes a few seconds after end of stimulation, with no net change of position. Attempts to reverse the paradigm, i.e. the claw is shocked if it closes beyond the designated threshold ('open-learning'), have not been successful so far (three failures, no successes). It may be possible, as in the locust 2'~, to use a paradigm in which the crayfish must learn to maintain its dactylopodite within a particular range. Although this report does not address the mechanisms underlying the change of claw position, we must in the future consider both central and peripheral contributions. A particularly relevant peripheral mechanism is the catch property that has been documented by Wilson and coworkers2O, 2~. They showed that a short, high frequency burst of stimulation could considerably enhance the tension produced by a continuing slow stimulation. In our conditions, the contraction of the positional claw closer muscle produced by an unchanging tonic motor nerve input could be enhanced for some minutes by the electrical stimulus. This known mechanism, however, does not explain the long duration of the effect during the test period nor the lack of position change exhibited by the control claw; thus, some central changes seem highly likely. The prospect of recording from many of the relevant neurons while an animal is learning remains an exciting one. This preparation offers a promising model system for studying the cellular basis of learning in an intact, behaving animal.
190 This work was supported by U S P H S G r a n t NS 05606. The a u t h o r s would like to t h a n k Dr. Bruce Lindsey for suggestions and assistance.
1 Atwood, H. L. and Wiersma, C. A. G., Command interneurons in the crayfish central nervous system, J. exp. BioL, 46 (1967) 249-261. 2 Disterhoft, J. F., Nurnberger, J. and Coming, W. C., 'P-R' differences in intact cockroaches as a function of testing interval, Psychon. Sci., 12 (1968) 205-206. 3 Eisenstein, E. M. and Cohen, M. J., Learning in an isolated prothoracic insect ganglion, Animal Behav., 13 (1965) 104-108. 4 Gelperin, A., Rapid food aversion learning by a terrestrial mollusk, Science, 189 (1975) 567--570. 5 Gerstein, G. L., Schuster, R. H. and Wiens, T. J., Neural activity and behavioral plasticity in the crayfish claw. In J. A. Kiger (Ed.), The Biology of Behavior, Oregon State Univ. Press, Corvallis, 1972, pp. 67-82. 6 Horridge, G. A., Learning of leg position by the ventral nerve cord in headless insects. Proc. Roy. Soc. B, 157 (1962) 33-52. 7 Horridge, G. A., Learning of leg position by headless insects, Nature (Lond.), 193 (1962) 697-698. 8 Horridge, G. A., The electrophysiological approach to learning in isolatable ganglia. In W. H. Thorpe and D. Davenport (Eds.), Animal Behaviour, SuppL 1, 1965. 9 Hoyle, G., Neurophysiologieal studies of learning in headless insects. In J. E. Treherne and J. W. L. Beament (Eds.), The Physiology of the lnsect Central Nervous System, Academic Press, London, 1965. 10 Kandel, E. R., An invertebrate system for the cellular analysis of simple behaviors. In F. O. Schmitt and F. G. Worden (Eds.), The Neurosciences: Third Study Program, MIT Press, Cambridge, 1974, pp. 347-370. 11 Lindsey, B. G. and Gerstein, G. L., Spatio-temporal characterization of crayfish claw efferents' firing patterns during step movements of the dactylopodite, Abstr., Soc. Neurosci., 6 (1976) 330. 12 Mpitsos, G. J. and Collins, S. D., Learning: rapid aversive conditioning in the gastropod mollusk Pleurobranchaea, Science, 188 (1975) 954-957. 13 Mpitsos, G. J. and Davis, W. J., Learning: classical and avoidance conditioning in the mollusk Pleurobranchaea, Science, 180 (1973) 317-320. 14 Pritchatt, D., Avoidance of electric shock by the cockroach Periplaneta americana, Anita Behav. 16 (1968) 178-185. 15 Smith, D. O., Central nervous control of excitatory and inhibitory neurons of the opener muscle of the crayfish claw, J. NeurophysioL, 37 (1974) 108-118. 16 Stafstrom, C. E. and Gerstein, G. L., A paradigm for position learining in the crayfish claw, Abstr., Soc. Neurosci., 2 (1976) 437. 17 Wiens, T. J. and Atwood, H. L., Intracellular studies of crayfish claw motoneurons, Abstr., Soc. Neurosci., 2 (1976) 260. 18 Wiens, T. J. and Gerstein G. L., Cross connections among crayfish claw efferents, .l. NeurophysioL, 38 (1975) 909-921. 19 Wiens, T. J. and Gerstein, G. L., Reflex pathways in the crayfish claw, J. ~mp. PhvsioL, At07 (1976) 309-326. 20 Wilson, D. M. and Larimer, J. L., The catch property of ordinary muscle, Proc. nat. Acad. Sci. (Wash.), 61 (1968) 909-916. 21 Wilson, D. M., Smith, D. O. and Dempster, P., Length and tension hysteresis during sinusoidal and step function stimulation of arthropod muscle, Amer. J. PhysioL, 218 (1970) 916-922: 22 Woollacott, M. H. and Hoyle, G., Membrane resistance changes associated with single, identified neuron learning, Abstr., Soc. NeuroscL 6 (1976) 339. 23 Woollacott, M. H., Tosney, T. and Hoyle, G., Pacemaker changes as a principal basis tbr insect leg learning, Abstr. Soc. Neurosci., 1 (1975) 501.