- Email: [email protected]
Midbrain unit activity during classical conditioning
Brain Research, 134 (1977) 73-82 ~) Elsevier/North-Holland Biomedical Press
73
M I D B R A I N U N I T A C T I V I T Y D U R I N G CLASSICAL C O N D I T I O N I N G
STEVEN E. BRAUTH* and JAMES OLDS Division of Biology, California Institute of Technology, Pasadena, CaliJ~ 91125 (U.S.A.)
(Accepted January 7th, 1977)
SUMMARY In behaving rats, unit activity recorded from the ventral tegmentum and from the reticular formation was monitored during classical conditioning. Rewarding electric stimulation of the medial forebrain bundle was used as the unconditioned stimulus (UCS). Only those cells possessing prior responses to the conditioned stimulus (CS) changed their responses as a result of conditioning. Responses recorded from cells which were driven by both the auditory CS and the brain shock UCS were significantly more often changed than those driven by the CS alone. These data show that the auditory and brain shock fields of influence interact in at least some brain regions prior to conditioning and that pairing the two kinds of stimuli is more likely to influence auditory responses recorded from these regions than those recorded elsewhere. It is possible to imagine that the intersection of the two fields is a sine qua non of conditioning and that the two prior actions caused the change by interacting at or near the recording point.
INTRODUCTION Many theories of conditioning and associative learning postulate the existence of neuron pools at "intersections" between CS and UCS message pathways in the brain 1,4,'~. The influences of the two signals must converge in space and time at some point in order for an association to take place. New responses to the CS might well emerge in brain regions outside the areas of intersection, but these would have to be secondary to other new responses which emerged withii~ the intersection field. It seems likely, therefore, that a priori connections with either the CS or the UCS would increase the chances of a neuron participating in the conditioning process. In this study neurons were monitored in the midbrain ventral tegmentum and * Present address: Department of Psychology, University of Maryland, College Park, Md. 20742 (U.S.A.).
74 reticular formation during a conditioning experiment. Many units in these regions show unconditioned responses to auditory and visual stimuli. Moreover, Kornblith and Olds 6 have shown that cells in the midbrain reticular formation evidence enhanced responding to an auditory tone CS after it has been paired with a food reward. Nevertheless, food reward is a complex stimulus whose neural mechanisms are nol understood well enough to allow identification of those cells driven by the reinforcement. Rewarding electrical stimulation of the medial forebrain bundle, however, does drive neurons throughout the midbrain and ventral forebrain which can be studied in a freely moving animalL For these reasons medial forebrain bundle stimulation was used as the UCS in the present experiment, and an auditory tone was used as the CS l- • The purpose of the experiment was to assess the extent to which a priori responding of midbrain units to either the auditory tone C S or brain stimulation reward altered the probability that such units would participate in the classical conditioning process. METHODS Fifteen male albino rats were used as subjects. These rats were pretrained to press a pedal for a 0.45 sec train of brain stimulation (0.3 msec pulses, 40 pulses/sec, monophasic). Brain stimulation (BRS) voltage levels were set from 5 to 8 V depending on the level needed to support self-stimulation in the animals. Six microelectroOes (62.5 # m diameter; enamel-insulated) were implanted in each animal. Methods of recording unit activity have been described previously 8. Spike height and time constant discriminators were used in this study and computergenerated "quality control" pictures of wave shape were obtained as well as signal-tonoise measures for all cell discharges included in the experiment. Ninety units passed the initial screening which consisted of signal-to-noise ratios of 4:1 and steady firing throughout the experiment. Typical unit responses are shown in the top frame of Fig. 2. The experiment was conducted on two successive days. During the first day delivery of the UCS (BRS) and the two CS stimuli (auditory tones; C S i 1000 Hz and CS- z 10,000 Hz or vice versa) were random (pseudoconditioning) with an average interstimulus interval of I rain. Each stimulus was presented 240 times in a 12 h period. On the second day of training pseudoconditioning was followed by conditioning. Each stimulus was presented randomly, as on day 1, for 60 trials; then conditioning began and one of the tones ( C S + ) was followed by the UCS after a 1 sec interval. The other tone (CS-) was presented on a random schedule of its own; 180 presentations of each stimulus were made during the conditioning. In addition to receiving a train of electrical pulses as the UCS, animals were also given an opportunity to bar-press for more brain stimulation. Immediately after the UCS train, a movable bar was activated and rats could self-stimulate for an additional 15 sec before the bar was retracted. This served to prove that stimulation was rewarding throughout the entire experimental procedure and all animals in the BRS group did sustain high levels of self-stimulation during the course of the experiment.
75
000
{13
~'{.0
I
1
I I I~
'~1- uO Eb
I
I
I
I
--LO T
b.J
O0
L
[
I
cOI --
I P--
i
I
I
~Or¢~ i
I',---
o_,
~
[
_
tt~r~')
3,~ o~
oc~ O~ J ea~o
ea~
L
L
d
_
0~'
_
_
d d "u!q aw!l 0 6u!Jnp ~!ds ~!un 0 ~o FI!I! m
Fig. 1. Graphs of the responses of individual units which showed statistically significant changes in responsiveness to the C S + as a function of conditioning. Each curve represents the average for a block of 60 trials. It is a peristimulus histogram. Each point represents the average probability of a unit detection in a 1.8 msec time bin averaged for a particular set of bins and for the 60 trials. These can be converted into an approximation of the firing rate (spikes/sec) by multiplying by 556 (because there were 556 bins/sec). Each point on the abscissa (except the first) represents 2 bins (3.6 msec). The time marks indicate time in msec from the arrival of sound at the ear to the end of the bin set. The first block of 60 trials is represented by a smooth curve (without symbols). The second 60 trials are represented by trianges; and the third 60 trials of the experimental session by squares. The first point on the abscissa represents the average spike probability during the 100 background bins (see text).
76
Fig. 2. Top: midbrain unit activity. Uppcr trace represents the output of spike height and time constant window discriminator. 100 msec sweep. Middle: reticular formation units driven by medial "forebrain bundle pulses (40/sec, 0.3 msec pulses). 100 msec sweep. Lower: same as middle, except 20 msec sweep. Two discriminable units are driven by the pulse.
Screening procedure]'or unit changes Unit responsiveness to the CS + was determined for each unit on an individual basis. This was done by constructing a two-factor analysis of variance for each unit, for each experimental session (both for pseudoconditioning and conditioning). T h e twofactor analysis o f variance was based on two main effects: the effect o f trials (in other words, the effect o f progression o f trials during the course o f each experimental session), and the effect o f time bins. In these experiments, unit response data were expressed as poststimulus histograms. Each time bin in the analysis represented 1.8 msec. In the analysis o f variance, therefore, the effect o f bins on unit activity was used to assess the extent to which the cell was responsive to the presentation of the CS +.
77
-STIMULATING ELECTRODE PLACEMENT
A 1.76 mm A 1 . 9 5 mm
Fig. 3. Locations of units studied in this experiment. Circled points and numbers represent CS driven units showing learning changes. Points numbered 1-11 represent locations of units driven by MFB UCS. All except number 11 were driven by the auditory CS as well as the brain shock USC. Driving latencies from the MFB were 1 = 2 msec, 2 = 4-10 msec, 3 -- 2-10 msec, 4 = 12 msec, 5 -- 2 msec, 6 = 2 msec, 7 = 4 msec, 8 = 8 msec, 9 -- 4-12 msec, 10 = 4 msec, 11 = 8 msec.
T h e effect o f trials w a s u s e d t o assess t h e e x t e n t t o w h i c h o v e r a l l a c t i v i t y c h a n g e d during the course o f each experimental session. In
order to carry
o u t t h e a n a l y s i s o f v a r i a n c e , t h e 25 b a c k g r o u n d
bins
i m m e d i a t e l y p r e c e d i n g e a c h p r e s e n t a t i o n o f t h e C S s t i m u l i a n d t h e 25 b i n s i m m e d i -
78 ately following the presentation of the CS stimuli were utilized. In order to eliminate any artifacts in those bins associated with the actual activation of the loudspeaker producing the CS stimuli, the first two bins immediately following the C S + (3.6 msec) were discarded and the next 25 bins (representing unit responses with latencies of 3.6-4.5 msec were utilized in the analysis). Since the air travel time of the auditory stimuli to the animal's head was determined to be 3 msec, this procedure did not result in the loss of any information concerning the nature of the responses of the cells studied to the CS stimuli. In addition to obtaining the above main effects due to trials and bins, the analysis of variance procedure allowed the computation of an interaction term. This interaction represents a bin effect change (response change to CS stimuli) not accounted for by overall changes in unit activity across the experimental trial sequence. It was this interaction term, therefore, which was taken as a measure of learning. In order to be considered as having changed response as a function of the conditioning procedure this interaction term was required to be significant at the 0.01 level for the conditioning session, but not for the pseudoconditioning session. Some units show changes in responsiveness to the CS stimuli due to habituation or sensitization processes. Such cells were not counted as having changed response as a function of the conditioning procedure. Thus each unit studied in this experiment was scored on the basis of whether it possessed innate responsiveness to the CS stimuli, and whether the responsiveness of the unit to the CS stimuli changed as a function of the conditioning procedure, using these statistical methods. A plot of the behavior each unit scored by this procedure to have changed response as a function of conditioning is provided ira Fig. I. RESULTS Although 90 units were studied in this experiment, only 10 showed changes in response to the C S + which were defined as learning changes by the above criteria. In
MIDBRAIN UNIT RESPONSES BEFORE ~ AFTER CONDtTIONING RF AFTER
/ ~ , , ~
~-VTA AVTAAFTER
_z
MSEC AFTER CS+ ONSET
Fig. 4. Composite graphs of unit responding in VTA and RF units. Units were pooled by normalizing response strength at each latency as a percent of background rate.
79 UNIT RESPONSECHANGESDURINGCONDITIONING (RESPONSE LATENCY12-16MSEC) w 601
,.
m0 c o ~ 40z~ Oc) ~m w 30L0 z ~ 2O
z
0
i
TRIAL GROUP
Fig. 5. Composite unit responses generated as in Fig. 3 plotted for response latencies of 12-16 msec across trial groups. the BRS group, 11 units were found which showed driving from the MFB stimulating electrode. Latencies ranged from 2 to 12 msec. In 3 cases the units showed variable latency in which responding to the brain stimulus pulses was confined to an interval following each pulse. Fig. 2 provides an example of this type of driving in the lower two frames. A map of the location of units evidencing driving and/or changes in responding due to conditioning is provided in Fig. 3. The R F units were located in the midtegmentum of the midbrain at points ranging from levels just rostral to those just caudal to the decussation of the superior cerebellar peduncle. VTA cells were found throughout the ventral portion of the rostral midbrain and caudal diencephalon at levels ranging from the most anterior extent of the red nucleus to the most posterior portions of the medial geniculate nucleus. The results of the analysis of those cells showing significant changes in responding to the C S + as a result of conditioning are presented in Fig. 4. The composite graphs indicate that neurons in the reticular formation begin to respond to the C S + at latencies of 4-8 msec; those in ventral tegmentum are recruited at longer latencies of 8-12 msec. For both regions, conditioning produces general enhancement of the response at latencies of 8-32 msec, with the greatest proportional changes occuring at latencies of 12-16 msec. Responding during this time interval is plotted as a function of the trial sequence in Fig. 5. Each trial group represents 10 trials. The first 6 groups of trials represent pseudoconditioning. VTA units show an early increase in response by group 8. This means that cells in this area show changes due to learning after 20 trials of the conditioning experiment have been presented. RF units, however, do not begin to change response strength until group 9, indicating that they participate in the conditioning process at a later stage. The number of cases was too small to make the difference between the two groups significant; nevertheless it is consistent with prior work 7. The expressed purpose of ,this study was to examine the behavior of those neurons driven by both CS and UCS stimuli. In order to do this each unit was scored on the basis of whether it showed prior responses to the C S + and/or the UCS and
80 TABLE I Units showing altered responding to the CS ~ as a result of conditioning Fractions represent numbers of units changing response to the CS + out of total in each group, i.e.,
2/44 indicates 2 units changed responses out of 44 in the group. Brain region
Proportion evidencing response changes due to conditioning
2/44 8/46
Ventral tegmentum Reticular formation
10/90
Totals
whether it changed response due to conditioning as defined by the criteria given in the Methods section of this paper. Units showing a significant response to the auditory tone CS + were found throughout the midbrain. A tally of the proportion of these cells showing conditioning response changes, as welt as the proportion of cells driven by the UCS showing response changes, is given in Tables I and II. These results indicate that neurons not possessing unconditioned responses to the CS + did not change response as a result of training (P < 0.01). In addition, units driven by both stimuli were more likely to condition than those driven by the CS alone (P < 0.05). Significantly more units in the reticular formation than in the ventral tegmentum (P < 0.05) changed response to the CSq-. This difference is probably due to the greater proportion of CS driven units in the reticular formation. DISCUSSION
The results of the experiment indicate that it is only those cells possessing prior connections to the C S pathway that show response changes. This is true in both reticular formation and ventral tegmentum. This implies that although the behavioral response of the animal arises de novo as a result of learning, only those midbrain units that possess connections to the CS pathway participate in the conditioning process. This effect constitutes strong evidence in favor of a model of learning based on the intersection of CS and UCS pathways. TABLE II Effects of a priori connections with CS + and UCS on unit conditioning in B R S animals Brain region
Type of driving
No driving
UCS Driving only
CS--, Driving only
CS + [ UCS Both
Totals
Ventral tegmentum Reticular formation
0/28 0/16
0/1 0/0
0/12 5/23
2/3 3/7
2/44 8/46
Subtotals
0/44
0/1
5/35
5/10
10/90
81 Further evidence for this theory is provided by a study of 11 units in this experiment which were actually driven by the electrical stimulus pulses in the BRS group. All but one of these units were also driven by the C S + . As shown in Table II, 50 % of the units driven by both stimuli showed changes in responding to the C S + as a result of conditioning, while only 14 % of the units driven exclusively by the auditory CS changed response during the conditioning procedure (P < 0.05). These results imply that only those midbrain units with prior connections to the C S + change response due to conditioning, and that neurons with prior connections to both CS and UCS pathways are significantly more likely to participate in the appetitive conditioning process. Additional support for this interpretation can be obtained by examining the behavior of 34 CS-driven units in the BRS animals which were accelerated, but not actually driven by the UCS with a fixed latency. Only 12 % of these neurons showed response changes, indicating that it is direct connection with the UCS pathway which enhances the probability that a cell will participate in the conditioning process. Very recently, Vertes and Miller 9 demonstrated that neurons in the pontine reticular formation participate in the aversive conditioning process. Neurons were found whose responses were accelerated to a conditioning stimulus after it was repeatedly paired with a footshock UCS. In addition, these changes in unit responsiveness did not appear to be the result of general arousal, insofar as other unconditioned stimuli, such as water, did not produce these effects. Vertes and Miller therefore concluded that these conditioned unit response changes were specific to the aversive conditioning procedure employed. The presently reported results are consistent with the above described findings. In the present experiment, neurons directly driven by the appetitive brain shock UCS were most likely to change response to a CS as a result of conditioning. Therefore in both the present experiment and the Vertes and Miller 9 experiment, brain stem neuronal response changes to a conditioning stimulus appeared to be specific to the kind of stimulus used as the UCS. This may mean that for both aversive and appetitive conditioning processes learning is dependent on neurons with prior connections to both CS and UCS stimuli. ACKNOWLEDGEMENT This research was supported by Public Health Service Grant MH-16978.
REFERENCES 1 Hebb, D. O., The Organization of Behavior: a Neuropsychologieal Theory, Wiley, New York, 1949. 2 lto, M., Excitability of medial forebrain bundle neurons during self-stimulating behavior, J. NeurophysioL, 35 (1972) 652-664. 3 lto, M. and Olds, J., Unit activity during self-stimulation behavior, J. NeurophysioL, 34 (1971) 263 273. 4 Kandel, E. R. and Tauc, L., Heterosynaptic facilitation in neurons of the abdominal ganglion of Aplysia depilans, J. Physiol. (Lond.), 181 (1965) 1-27.
82 5 Kandel, E. R. and Tauc, L., Mechanism of heterosynaptic facilitation in the giant cell of the abdominal ganglion of Aplysia depilans, J. Physiol. (Lond.), 181 (1965) 28-47. 6 Kornblith, C. and Olds, J., Unit activity in brain stem reticular formation of the rat during learning, J. Neurophysiol., 36 (1973) 489-501. 7 0 l d s , J., Brain mechanisms in reinforcement learning. In Pleasure, Reward, Preference, Academic Press, New York, 1973, pp. 35-63. 8 0 l d s , J., Disterhoft, J. F., Segal, M., Kornblith, C. L. and Hirsh, R., Learning centers of the rat brain mapped by measuring latencies of conditioned unit responses, J. Neurophysiol., 35 (1972) 202-219. 9 Vertes, R. P. and Miller, N. E., Brain stem neurons that fire selectively to a conditioned stimulus for shock, Brain Research, 103 (1976) 229-242.
73
M I D B R A I N U N I T A C T I V I T Y D U R I N G CLASSICAL C O N D I T I O N I N G
STEVEN E. BRAUTH* and JAMES OLDS Division of Biology, California Institute of Technology, Pasadena, CaliJ~ 91125 (U.S.A.)
(Accepted January 7th, 1977)
SUMMARY In behaving rats, unit activity recorded from the ventral tegmentum and from the reticular formation was monitored during classical conditioning. Rewarding electric stimulation of the medial forebrain bundle was used as the unconditioned stimulus (UCS). Only those cells possessing prior responses to the conditioned stimulus (CS) changed their responses as a result of conditioning. Responses recorded from cells which were driven by both the auditory CS and the brain shock UCS were significantly more often changed than those driven by the CS alone. These data show that the auditory and brain shock fields of influence interact in at least some brain regions prior to conditioning and that pairing the two kinds of stimuli is more likely to influence auditory responses recorded from these regions than those recorded elsewhere. It is possible to imagine that the intersection of the two fields is a sine qua non of conditioning and that the two prior actions caused the change by interacting at or near the recording point.
INTRODUCTION Many theories of conditioning and associative learning postulate the existence of neuron pools at "intersections" between CS and UCS message pathways in the brain 1,4,'~. The influences of the two signals must converge in space and time at some point in order for an association to take place. New responses to the CS might well emerge in brain regions outside the areas of intersection, but these would have to be secondary to other new responses which emerged withii~ the intersection field. It seems likely, therefore, that a priori connections with either the CS or the UCS would increase the chances of a neuron participating in the conditioning process. In this study neurons were monitored in the midbrain ventral tegmentum and * Present address: Department of Psychology, University of Maryland, College Park, Md. 20742 (U.S.A.).
74 reticular formation during a conditioning experiment. Many units in these regions show unconditioned responses to auditory and visual stimuli. Moreover, Kornblith and Olds 6 have shown that cells in the midbrain reticular formation evidence enhanced responding to an auditory tone CS after it has been paired with a food reward. Nevertheless, food reward is a complex stimulus whose neural mechanisms are nol understood well enough to allow identification of those cells driven by the reinforcement. Rewarding electrical stimulation of the medial forebrain bundle, however, does drive neurons throughout the midbrain and ventral forebrain which can be studied in a freely moving animalL For these reasons medial forebrain bundle stimulation was used as the UCS in the present experiment, and an auditory tone was used as the CS l- • The purpose of the experiment was to assess the extent to which a priori responding of midbrain units to either the auditory tone C S or brain stimulation reward altered the probability that such units would participate in the classical conditioning process. METHODS Fifteen male albino rats were used as subjects. These rats were pretrained to press a pedal for a 0.45 sec train of brain stimulation (0.3 msec pulses, 40 pulses/sec, monophasic). Brain stimulation (BRS) voltage levels were set from 5 to 8 V depending on the level needed to support self-stimulation in the animals. Six microelectroOes (62.5 # m diameter; enamel-insulated) were implanted in each animal. Methods of recording unit activity have been described previously 8. Spike height and time constant discriminators were used in this study and computergenerated "quality control" pictures of wave shape were obtained as well as signal-tonoise measures for all cell discharges included in the experiment. Ninety units passed the initial screening which consisted of signal-to-noise ratios of 4:1 and steady firing throughout the experiment. Typical unit responses are shown in the top frame of Fig. 2. The experiment was conducted on two successive days. During the first day delivery of the UCS (BRS) and the two CS stimuli (auditory tones; C S i 1000 Hz and CS- z 10,000 Hz or vice versa) were random (pseudoconditioning) with an average interstimulus interval of I rain. Each stimulus was presented 240 times in a 12 h period. On the second day of training pseudoconditioning was followed by conditioning. Each stimulus was presented randomly, as on day 1, for 60 trials; then conditioning began and one of the tones ( C S + ) was followed by the UCS after a 1 sec interval. The other tone (CS-) was presented on a random schedule of its own; 180 presentations of each stimulus were made during the conditioning. In addition to receiving a train of electrical pulses as the UCS, animals were also given an opportunity to bar-press for more brain stimulation. Immediately after the UCS train, a movable bar was activated and rats could self-stimulate for an additional 15 sec before the bar was retracted. This served to prove that stimulation was rewarding throughout the entire experimental procedure and all animals in the BRS group did sustain high levels of self-stimulation during the course of the experiment.
75
000
{13
~'{.0
I
1
I I I~
'~1- uO Eb
I
I
I
I
--LO T
b.J
O0
L
[
I
cOI --
I P--
i
I
I
~Or¢~ i
I',---
o_,
~
[
_
tt~r~')
3,~ o~
oc~ O~ J ea~o
ea~
L
L
d
_
0~'
_
_
d d "u!q aw!l 0 6u!Jnp ~!ds ~!un 0 ~o FI!I! m
Fig. 1. Graphs of the responses of individual units which showed statistically significant changes in responsiveness to the C S + as a function of conditioning. Each curve represents the average for a block of 60 trials. It is a peristimulus histogram. Each point represents the average probability of a unit detection in a 1.8 msec time bin averaged for a particular set of bins and for the 60 trials. These can be converted into an approximation of the firing rate (spikes/sec) by multiplying by 556 (because there were 556 bins/sec). Each point on the abscissa (except the first) represents 2 bins (3.6 msec). The time marks indicate time in msec from the arrival of sound at the ear to the end of the bin set. The first block of 60 trials is represented by a smooth curve (without symbols). The second 60 trials are represented by trianges; and the third 60 trials of the experimental session by squares. The first point on the abscissa represents the average spike probability during the 100 background bins (see text).
76
Fig. 2. Top: midbrain unit activity. Uppcr trace represents the output of spike height and time constant window discriminator. 100 msec sweep. Middle: reticular formation units driven by medial "forebrain bundle pulses (40/sec, 0.3 msec pulses). 100 msec sweep. Lower: same as middle, except 20 msec sweep. Two discriminable units are driven by the pulse.
Screening procedure]'or unit changes Unit responsiveness to the CS + was determined for each unit on an individual basis. This was done by constructing a two-factor analysis of variance for each unit, for each experimental session (both for pseudoconditioning and conditioning). T h e twofactor analysis o f variance was based on two main effects: the effect o f trials (in other words, the effect o f progression o f trials during the course o f each experimental session), and the effect o f time bins. In these experiments, unit response data were expressed as poststimulus histograms. Each time bin in the analysis represented 1.8 msec. In the analysis o f variance, therefore, the effect o f bins on unit activity was used to assess the extent to which the cell was responsive to the presentation of the CS +.
77
-STIMULATING ELECTRODE PLACEMENT
A 1.76 mm A 1 . 9 5 mm
Fig. 3. Locations of units studied in this experiment. Circled points and numbers represent CS driven units showing learning changes. Points numbered 1-11 represent locations of units driven by MFB UCS. All except number 11 were driven by the auditory CS as well as the brain shock USC. Driving latencies from the MFB were 1 = 2 msec, 2 = 4-10 msec, 3 -- 2-10 msec, 4 = 12 msec, 5 -- 2 msec, 6 = 2 msec, 7 = 4 msec, 8 = 8 msec, 9 -- 4-12 msec, 10 = 4 msec, 11 = 8 msec.
T h e effect o f trials w a s u s e d t o assess t h e e x t e n t t o w h i c h o v e r a l l a c t i v i t y c h a n g e d during the course o f each experimental session. In
order to carry
o u t t h e a n a l y s i s o f v a r i a n c e , t h e 25 b a c k g r o u n d
bins
i m m e d i a t e l y p r e c e d i n g e a c h p r e s e n t a t i o n o f t h e C S s t i m u l i a n d t h e 25 b i n s i m m e d i -
78 ately following the presentation of the CS stimuli were utilized. In order to eliminate any artifacts in those bins associated with the actual activation of the loudspeaker producing the CS stimuli, the first two bins immediately following the C S + (3.6 msec) were discarded and the next 25 bins (representing unit responses with latencies of 3.6-4.5 msec were utilized in the analysis). Since the air travel time of the auditory stimuli to the animal's head was determined to be 3 msec, this procedure did not result in the loss of any information concerning the nature of the responses of the cells studied to the CS stimuli. In addition to obtaining the above main effects due to trials and bins, the analysis of variance procedure allowed the computation of an interaction term. This interaction represents a bin effect change (response change to CS stimuli) not accounted for by overall changes in unit activity across the experimental trial sequence. It was this interaction term, therefore, which was taken as a measure of learning. In order to be considered as having changed response as a function of the conditioning procedure this interaction term was required to be significant at the 0.01 level for the conditioning session, but not for the pseudoconditioning session. Some units show changes in responsiveness to the CS stimuli due to habituation or sensitization processes. Such cells were not counted as having changed response as a function of the conditioning procedure. Thus each unit studied in this experiment was scored on the basis of whether it possessed innate responsiveness to the CS stimuli, and whether the responsiveness of the unit to the CS stimuli changed as a function of the conditioning procedure, using these statistical methods. A plot of the behavior each unit scored by this procedure to have changed response as a function of conditioning is provided ira Fig. I. RESULTS Although 90 units were studied in this experiment, only 10 showed changes in response to the C S + which were defined as learning changes by the above criteria. In
MIDBRAIN UNIT RESPONSES BEFORE ~ AFTER CONDtTIONING RF AFTER
/ ~ , , ~
~-VTA AVTAAFTER
_z
MSEC AFTER CS+ ONSET
Fig. 4. Composite graphs of unit responding in VTA and RF units. Units were pooled by normalizing response strength at each latency as a percent of background rate.
79 UNIT RESPONSECHANGESDURINGCONDITIONING (RESPONSE LATENCY12-16MSEC) w 601
,.
m0 c o ~ 40z~ Oc) ~m w 30L0 z ~ 2O
z
0
i
TRIAL GROUP
Fig. 5. Composite unit responses generated as in Fig. 3 plotted for response latencies of 12-16 msec across trial groups. the BRS group, 11 units were found which showed driving from the MFB stimulating electrode. Latencies ranged from 2 to 12 msec. In 3 cases the units showed variable latency in which responding to the brain stimulus pulses was confined to an interval following each pulse. Fig. 2 provides an example of this type of driving in the lower two frames. A map of the location of units evidencing driving and/or changes in responding due to conditioning is provided in Fig. 3. The R F units were located in the midtegmentum of the midbrain at points ranging from levels just rostral to those just caudal to the decussation of the superior cerebellar peduncle. VTA cells were found throughout the ventral portion of the rostral midbrain and caudal diencephalon at levels ranging from the most anterior extent of the red nucleus to the most posterior portions of the medial geniculate nucleus. The results of the analysis of those cells showing significant changes in responding to the C S + as a result of conditioning are presented in Fig. 4. The composite graphs indicate that neurons in the reticular formation begin to respond to the C S + at latencies of 4-8 msec; those in ventral tegmentum are recruited at longer latencies of 8-12 msec. For both regions, conditioning produces general enhancement of the response at latencies of 8-32 msec, with the greatest proportional changes occuring at latencies of 12-16 msec. Responding during this time interval is plotted as a function of the trial sequence in Fig. 5. Each trial group represents 10 trials. The first 6 groups of trials represent pseudoconditioning. VTA units show an early increase in response by group 8. This means that cells in this area show changes due to learning after 20 trials of the conditioning experiment have been presented. RF units, however, do not begin to change response strength until group 9, indicating that they participate in the conditioning process at a later stage. The number of cases was too small to make the difference between the two groups significant; nevertheless it is consistent with prior work 7. The expressed purpose of ,this study was to examine the behavior of those neurons driven by both CS and UCS stimuli. In order to do this each unit was scored on the basis of whether it showed prior responses to the C S + and/or the UCS and
80 TABLE I Units showing altered responding to the CS ~ as a result of conditioning Fractions represent numbers of units changing response to the CS + out of total in each group, i.e.,
2/44 indicates 2 units changed responses out of 44 in the group. Brain region
Proportion evidencing response changes due to conditioning
2/44 8/46
Ventral tegmentum Reticular formation
10/90
Totals
whether it changed response due to conditioning as defined by the criteria given in the Methods section of this paper. Units showing a significant response to the auditory tone CS + were found throughout the midbrain. A tally of the proportion of these cells showing conditioning response changes, as welt as the proportion of cells driven by the UCS showing response changes, is given in Tables I and II. These results indicate that neurons not possessing unconditioned responses to the CS + did not change response as a result of training (P < 0.01). In addition, units driven by both stimuli were more likely to condition than those driven by the CS alone (P < 0.05). Significantly more units in the reticular formation than in the ventral tegmentum (P < 0.05) changed response to the CSq-. This difference is probably due to the greater proportion of CS driven units in the reticular formation. DISCUSSION
The results of the experiment indicate that it is only those cells possessing prior connections to the C S pathway that show response changes. This is true in both reticular formation and ventral tegmentum. This implies that although the behavioral response of the animal arises de novo as a result of learning, only those midbrain units that possess connections to the CS pathway participate in the conditioning process. This effect constitutes strong evidence in favor of a model of learning based on the intersection of CS and UCS pathways. TABLE II Effects of a priori connections with CS + and UCS on unit conditioning in B R S animals Brain region
Type of driving
No driving
UCS Driving only
CS--, Driving only
CS + [ UCS Both
Totals
Ventral tegmentum Reticular formation
0/28 0/16
0/1 0/0
0/12 5/23
2/3 3/7
2/44 8/46
Subtotals
0/44
0/1
5/35
5/10
10/90
81 Further evidence for this theory is provided by a study of 11 units in this experiment which were actually driven by the electrical stimulus pulses in the BRS group. All but one of these units were also driven by the C S + . As shown in Table II, 50 % of the units driven by both stimuli showed changes in responding to the C S + as a result of conditioning, while only 14 % of the units driven exclusively by the auditory CS changed response during the conditioning procedure (P < 0.05). These results imply that only those midbrain units with prior connections to the C S + change response due to conditioning, and that neurons with prior connections to both CS and UCS pathways are significantly more likely to participate in the appetitive conditioning process. Additional support for this interpretation can be obtained by examining the behavior of 34 CS-driven units in the BRS animals which were accelerated, but not actually driven by the UCS with a fixed latency. Only 12 % of these neurons showed response changes, indicating that it is direct connection with the UCS pathway which enhances the probability that a cell will participate in the conditioning process. Very recently, Vertes and Miller 9 demonstrated that neurons in the pontine reticular formation participate in the aversive conditioning process. Neurons were found whose responses were accelerated to a conditioning stimulus after it was repeatedly paired with a footshock UCS. In addition, these changes in unit responsiveness did not appear to be the result of general arousal, insofar as other unconditioned stimuli, such as water, did not produce these effects. Vertes and Miller therefore concluded that these conditioned unit response changes were specific to the aversive conditioning procedure employed. The presently reported results are consistent with the above described findings. In the present experiment, neurons directly driven by the appetitive brain shock UCS were most likely to change response to a CS as a result of conditioning. Therefore in both the present experiment and the Vertes and Miller 9 experiment, brain stem neuronal response changes to a conditioning stimulus appeared to be specific to the kind of stimulus used as the UCS. This may mean that for both aversive and appetitive conditioning processes learning is dependent on neurons with prior connections to both CS and UCS stimuli. ACKNOWLEDGEMENT This research was supported by Public Health Service Grant MH-16978.
REFERENCES 1 Hebb, D. O., The Organization of Behavior: a Neuropsychologieal Theory, Wiley, New York, 1949. 2 lto, M., Excitability of medial forebrain bundle neurons during self-stimulating behavior, J. NeurophysioL, 35 (1972) 652-664. 3 lto, M. and Olds, J., Unit activity during self-stimulation behavior, J. NeurophysioL, 34 (1971) 263 273. 4 Kandel, E. R. and Tauc, L., Heterosynaptic facilitation in neurons of the abdominal ganglion of Aplysia depilans, J. Physiol. (Lond.), 181 (1965) 1-27.
82 5 Kandel, E. R. and Tauc, L., Mechanism of heterosynaptic facilitation in the giant cell of the abdominal ganglion of Aplysia depilans, J. Physiol. (Lond.), 181 (1965) 28-47. 6 Kornblith, C. and Olds, J., Unit activity in brain stem reticular formation of the rat during learning, J. Neurophysiol., 36 (1973) 489-501. 7 0 l d s , J., Brain mechanisms in reinforcement learning. In Pleasure, Reward, Preference, Academic Press, New York, 1973, pp. 35-63. 8 0 l d s , J., Disterhoft, J. F., Segal, M., Kornblith, C. L. and Hirsh, R., Learning centers of the rat brain mapped by measuring latencies of conditioned unit responses, J. Neurophysiol., 35 (1972) 202-219. 9 Vertes, R. P. and Miller, N. E., Brain stem neurons that fire selectively to a conditioned stimulus for shock, Brain Research, 103 (1976) 229-242.