PsyCHOPHYSIOLOGY
H •
Copyright © 1987 by The Society for Psychophysioiogical Research, Inc.
IIc *
Pnnted in U.SA,
Presidential Address^ 1986 Behavioural Energetics: Some Effects d Uncertainty on the Mobilization and Distribution of Energy JASPER BRENER University of Hull. England
ABSTRACT Several sets of data illustrating dissociation of metabolic and muscular work rates are presented. During response acquisition, the rate of energy expenditure declines although work rates remain constant and goal achievement increases. When higher response costs are demanded for goal achievement, work rates increase whereas energy expenditure remains constant. Decreases in outcome probability also give rise to increases in motor intensity and work rate but do not influence the rate of energy expenditure. It is proposed tbat behaviourally-related variations in metabolic rate are influenced by environmental uncertainty ("what to do"), whereas variations in tbe intensity of motor output are influenced by response uncertainty ("bow to do it"). The processes of response selection wbicb provide the means by wbich response uncertainty is resolved, activate motor channels for the expression of energy mobilized by tbe prevailing state of environmental uncertainty. DESCRIPTORS; Energy expenditure. Motor learning. Uncertainty, Cardiovascular-striate muscular interaction. My interest in the subject of this lecture arose during research on behaviourally-related variations in heart rate (HR). The results of our experiments on that subject agree with those reported by many other investigators. In particular, a substantial proportion of behaviourally-related HR variance can be accounted for by metabolic variation. Where the experimental contingencies demand an explicit mo-
The work I shall report was done in collaboration with a number of my students and colleague^s. I ^^^J^^^ pamcu a r y 1 e o an ei p, Sam Conna^W and"su2annTMitchen. whoTresearch I have drawn on in preparing this lecture. Thanks are also due to our technical staff: Ray Wallis, Clive Welboume, Sandra Readhead, Bob Richman, and Mark Hutchinson, without whose expen assistance the work could not have been executed. Finally, I would like to thank my recent Special Optionists, Paul Chung, Sarah Dilks, Jim Freund, Heather Harper, George Hedley, Sarah Mossman, Stephanie Oakley, Christopher Ring, and Anne Rockett, who spent many hours with me, patiently discussing the matters raised here and running experiments. Address requests for repnnts to Jasper Brener, who is now at the Department of Psychology. State University of New York at Stony Brook, Stony Brook, New York 11794.
tor response, this proportion can be greater than 80% (e.g., Brener, Phillips, & Sherwood, 1983). As Obrist (1976) in particular has emphasized, this is to be expected on the basis of the cardiovascular system's primary biological role in perfusing the tissues at rates which are proportional to their rates of activity. The close covariation of metabolic and blood flow rates illustrated in Figure 1 emphasizes the prominence of the energetic functions of the cardiovascular system. There have been few if any convincing dem^^^^^^^^-^^^^ ^^^^ ^^^ functions of the heart may be influenced through pathways that are independent of those which support its linkage to metabolic processes. Although the mechanisms of this Imkage are not fully understood, it is clear that the Striate muscles are important mediators. This effector system exerts effects on metabolic rate and on cardiovascular performance that are both powerful and compj^^^ p^^. example, when metabolic rate is held con^^^^^ cardiac activity varies systematically as a ^^^^^.^^ ^^ ^^^ ^^^^^^^^ muscular activity: the . cc ^ j i , •. u e.u^^^ i^ ,u^ relative density of fast and slow witch fibers m the target muscles, whether the muscle masses involved are large or small, and whether the work they perform is static or dynamic (Lewis et al., 1983; Sten-
499
500
Brener
Vol. 24. No. 5
I shall start by showing evidence that psychological processes are associated with reliable variations in energy expenditure. Then I shall turn to the question of how these variations are produced with special reference to the involvement of striate muscular performance. The attribution of behavioural ly-re la ted variations in energy expenditure to variations in striate muscular activity is soundly based in physiology and biology. However, the expectation that variations is energy expenditure will be paralleled by variations in striate muscular aclZ 15 tivity is questioned by some experimental evidence. ENERGY EXPENDITURE (BHRs) Such evidence leads to an examination of behavFigure 1. Distribtition of cardiac output as a function ioural efficiency since this concept incorporates the of metabolic rate (after Rowell. 1974). idea that rates of work and energy expenditure may exhibit independent variation. Although it is widely believed that behaviour tends to maximal efficienberg, Astrand, Ekblom, Royce, & Saltin, 1967). Fur- cy, the available evidence is slender. Observations thermore, the same metabolic load may be gener- of the covariation of motor performance and energy ated by quite different patterns of activity, even in expenditure provide some information on this basthe same muscles. These considerations make it ic issue and also some clues about how programs possible to argue that the cardiovascular variations for increased efficiency might be implemented. reported by psychophysiologists are produced Behavioural Energetics through pathways that have evolved to support the metabolic functions of the circulation. This inPsychology and Energetics cludes so-called "metabolically-unwarranted" eleEnergetic concepts are found in all areas of psyvations in cardiac performance although in such cases, central pathways supporting these metabolic chology: psychodynamics, ethology, learning thefunctions may play a prominent role {Brener, 1986a; ory, and cognitive psychology. Frequently the psychological concept of energy is only metaphorically 1986b; Engel, 1986). As a point of departure for my lecture I wish to related to the biophysical concept. Thus, the conentertain the idea that "psychogenic" variations in structs of libido, action-specific energy, drive, cardiac performance are manifestations of the met- arousal, etc., are to greater or lesser extents expresabolic variations that accompany behaviour. This sions of the concept of "psychic energy" (Hinde, does not, in my view, make them any less inter- 1960; Van den Berg, 1986). However in psychoesting as psychophysiologieal phenomena. On the physiology, e.g. in the ideas of Cannon (1929), Dufcontrary, it would seem to increase the opportun- fy (1951), Freeman (1948), Gellhorn (1967), and ities for accounting in a convincing way for vari- Hess (1954), we find a closer isomorphism between ations in cardiac activity. At the same time, it sug- the behavioural and biophysical concepts. Thus, gests that in order to understand how environmen- "emergency reactions," "ergotropic responses," tal demands cause cardiac variation, one has to drop "behavioural energetics," and "energy mobilizadown one level in the causal chain and try to un- tions" refer directly to processes which govern the derstand how behavioural and energetic variations mobilization and distribution of bodily energy stores are related. It seems odd that although energetic for the commission of adaptive behaviour. This refconcepts are so prominent in psychological theory, erence defines a common ground shared by the subthere are so few reports in the literature of how ject matter of psychology and other biological disenergy expenditure varies with behaviour. How- ciplines and provides a link between many apparever, since in our HR experiments we have rou- ently diverse lines of behavioural and psychophystinely made continuous measurements of whole- iologieal investigation. body oxygen consumption, we were naturally led Overall Energy Expenditure to observe the covariation of energetic and behavioural variables. These observations, many of which Overall energy expenditure exhibits a number of I shall be reporting for the first time here, form the remarkable constancies. The average rates at which main substance of the lecture. All findings reported organisms bum energy is related to their body mashere for the first time are significant at the .05 level ses by a universal biological law expressed by Kleiunless otherwise indicated. ber's formula: 0 nUSCLE • UISCEHfl QHEfiRT,BRAIN,etc. • SKIN
Behavioural Energetics
September. 1987 ,;,4hr = 7 0 M k g " .
The metabolic rates of animals as different as the harvest mouse which weighs lOOg and the African elephant which weighs in excess of 3000kg fall on the same straight line described by this function. Although Kleiber (1961) derived the law empirically, Economos (1982) has argued that the .75 exponent is deducible from more fundamental biophysical principles: the law of body surface whieh recognizes that the rate of respiration is determined by the surface area of the cells, and the anti-gravity principle which recognizes that more work is required to move heavier masses. All the energy expenditure data I shall present are expressed in terms of BMR (Basal Metabolic Rate) units since this enables inter-species comparisons.' The overall rate of energy expenditure may vary by as much as 15-20 times in the short term. However, normal individuals in sedentary occupations bum energy at an average rate of between 1.4 and 1.7 BMRs (Astrand & Rodahl, 1977) depending on leisure-time activity rates. Heavy manual labor, on the other hand, may be sustained at costs of approximately 4.5 BMRs for 8 hrs. Most of the energy expenditure in excess of the BMR is attributed to striate muscle activity (see Figure 1). Since this effector system provides the sole means of acting upon the environment, there seem to be good logical and biological reasons for emphasizing its role in accounting for behaviourally-related variations in energy expenditure. The Relationship between Striate Muscular Activity and Energy Expenditure
501
tween one tenth and one quarter of their daily energy expenditure. While digestion and thermoregulation may also contribute substantially to variations in whole-body energy expenditure, the effects of these processes may be controlled in the experimental situation, thereby enabling the covariation of energy expenditure and striate muscular activity to be studied. By transforming bodily energy stores into kinetic energy which is then expressed as work, the striate muscles permit organisms to do commerce with their environments. All measurable work is performed by this effector system which comprises 50% of the total body mass. Despite its long-term stability, the rate of energy expenditure does exhibit well-defined behaviourally-related variations. Reliable circadian fluctuations are associated with daily work and maintenance rhythms. Superimposed on these daily rhythms are asynchronous changes related to meeting unpredictable challenges that may arise during the day. As illustrated by the data presented in Figure 2, even relatively minor challenges result in reliable alterations in energy expenditure. The top diagram shows the effects on metabolic rate of requiring subjects to solve anagrams (Brener & Meakin, 1980). Each bar represents the mean metabolic rate for 12 subjects recorded over 5-min periods while they rested (RESTl and REST2), solved anagrams (SOLVE), and read solutions (READ) from the computer screen. Solutions were read back during the READ period at the same rate that they were produced during the SOLVE period thereby controlling for the metabolic costs of vocal activity. The small but reliable elevation in metabolic rate is of approximately the same magnitude
Although striate muscular activity is not, of course, the exclusive determinant of metabolic rate, this effector system is by far the most conspicuous user of energy. In 3 min of heavy exercise the striate muscles consume enough energy to maintain the whole body in a resting state for an hour. At the other extreme, during 8 hrs of sleep in which the striate muscles are quiet, the individual burns be-
'BMR values expressed in kilocalories were estimated by Kleiber's formula on the basis of the subject's body weight. These estimates were then divided by 4.82 to yield O2 equivalents of the BMR. Recorded rates of O; consumption were then divided by these values to provide measures expressed as multiples of the subject's BMR. The advantage of expressing energy expenditure in relation to BMR rather than in relation to body weight is that the former method takes account of the systematic differences in metabolic rate that are associated with variations in body size.
REST
TUSK
Figure 2. Changes in metabolic rate produced by laboratory tasks in humans. Top Diagram; solving anagrams. Bottom Diagram: (a) stressful reaction time task; (b) playing a video game.
502
Brener
as that recorded by Benedict and Benedict (1933) under similar conditions. Those investigators argued that the increase in energy expenditure associated with mental work was produced by associated striate muscular activity. A similar conclusion is reached by Astrand and Rodahl {1977) who note that other investigators have observed increases in metabohc rate of 11% during mental work. These views are supported by failures to find evidence that the overall rate of mental activity is correlated with the metabolic rate of the brain (Van den Berg, 1986). The lower diagram in Figure 2 illustrates changes in energy expenditure in human subjects at rest, during a stressful RT task (Sherwood, Allen, Obrist, & Langer, 1986), and while playing a video game (Tumer & Carroll, 1985). It will be noted that both challenges give rise to increases in metabolic rate of about 30%. The low resting metabolic rates depicted here may be associated with the stringent physical restraints applied to subjects in these experiments. They were seated, discouraged from moving, and had a number of transducers attached to their heads and bodies. Nevertheless it will be noted that in these experiments where limb movements were required to satisfy task requirements, augmentations in energy expenditure were more substantial than those recorded in the anagram experiment. As to be expected, more profound energetic variations are associated with behavioural adjustments that implicate greater muscle masses. Figure 3 illustrates alterations in metabolic rates during shockavoidance behaviour in rats (Brener, Phillips, & Connally, 1980) and dogs (Langer, Obrist, & McCubbin, 1979), and in rats working for food (Brener et al., 1983). During task periods, the relatively unrestrained subjects in these experiments exhibited metabolic rates of about 3 BMRs. Differences in the REST rates of energy expenditure may be due to the use of a discrete rest period in the dog experiment and recycling S^/S-' periods in the rat experiments.
Vo!, 24. No. 5
RESPONSE SELECTION EFFECTORS ENERGY HOBILIZRTION Figure 4. Serial model of energy control: energy expenditure is secondary to the rate of effector activity.
All these task-related variations in energy expenditure may be explained in terms of a serial model in which the brain controls the muscles and muscles determine the rate of energy expenditure (Figure 4). In this model, the brain need not concern itself with controlling the rate of energy expenditure which is an automatic product of the rate of effector (particularly striate muscle) activity. However problems for tbe model arise in data such as those illustrated in Figure 5. Tbis figure illustrates ambulation and energy expenditure rates in rats that received food contingent on running (EXPERIMENTALs) and independently of behaviour (YOKED-CONTROLs). It will be seen that energy expenditure changes are correlated with ambulation changes in EXPERIMENTAL but not in YOKED-CONTROL animals.
i »
2,9
S
2,7
1EXPERIMENTDL S IVOKED-CDHTROL
I 2,6 2.5 l.l l.l 2,1
REST Figure 3. Changes in metabolic rate produced by laboratory tasks in animals, (a) rats avoiding shocks, (b) dogs avoiding shocks, and (c) rats running for food.
JI
REST TBSK " REST TflSK Figure 5. Ambulation and energy expenditure rates in rats running for food ( E X P E R I M E N T A L ) a n d rats receiving food independently of their behaviour {YOKEDC O N T R O L ) . Note that ambulation a n d energy expenditure rates are correlated in Experimental but not in YokedControl animals.
Seplcmber. 1987
503
Behavioural Energetics
Hence, iti the case of the YOKED-CONTROLs, energy expenditure cannot be attributed sitnply to the recorded rate of striate muscular work.
HIGH
PERfORtiANCE
The Covariation of Energy Expenditure and Task Work Rates
Several other sets of data oppose the notion that energy expenditure is a simple consequence of the rate of work. The idea that energetic processes may be governed relatively independently of motor control and activation processes is not uncommon in behavioural theory. For example, the expectation that performance efficiency will increase with practice, anticipates a dissociation between work and energy expenditure rates. In work physiology, the energetic efficiency of performance is generally expressed by the ratio of the energy requirements of the work to the energy consumed by the behaviour in performing the work {Banister & Brown, 1968). When work and energy expenditure are expressed in the same units, the maximum efficiency for striate muscular activity is calculated to be in the region of 25%. However, behavioural efficiency is usually far less. In order for it to increase, work output must be augmented relatively more than energy expenditure. Although there is a ijervasive belief in the behavioural sciences that the energy efficiency of behaviour does increase with training, it is based on rather indirect evidence. Therefore we undertook a series of experiments in which energy expenditure and a broad range of kinetic and temporal performance variables were monitored continuously. Our analyses of behavioural efficiency are based on similar assumptions to those employed by Norberg (1977) in his theoretical treatment of foraging in natural situations. In particular, the whole-body energy expenditure is seen to be comprised of three principal elements: a) a basal metabolic provision, b) a provision for task-related behaviour, and c) a provision for task-irrelevant behaviour. Whereas the BMR requirement will remain fixed, the provision for task-irrelevant behaviour may decrease without compromising task performance. Furthermore, the energy costs of task-related behaviour may also decrease due to refinements of effective behavioural variants. Thus increases in behavioural efficiency may be achieved by eliminating activities that are not required by the task. Presumably such processes account for the observations made by work and exercise physiologists that with practice, less energy is expended per unit of work performed (e.g. Stegemann, 1981). Such observations conform to the theoretical relationship between energy expenditure and performance de-
ENER6Y EXPEHDITURE
LOU
TRAINING Figure 6. Hypothesized alterations in energy expenditure during skill learning (after Freeman, 1948, and Sparrow, 1983).
scribed by Freeman (1948) and Sparrow (1983) and illustrated in Figure 6. Where energy expenditure and work rates have been recorded in behavioural experiments, similar observations have been made. An example is provided in Figure 7 which depicts increases in the Operant and Energy Efficiency of performance as rats learned an ambulatory avoidance response.- It will be noted that in this example, efficiency has been expressed in terms of the amount of energy expended or work performed to achieve the task goal. This is because in the characterization of adaptive behaviour, energy expended and work performed per goal unit, seems a more appropriate metric of efficiency than energy expended per work unit. A similar analysis has been proposed by Martin and Levey (1965) in the analysis of conditioned eyeblink responses. They suggest that a "perfect" conditioned eyeblink ".. . is one which completely avoids the UCS, making the UCR unnecessary" (page 781). In an effort to examine how animals might reorganize their performance so as to earn reinforcement at lower cost, we have recently conducted a
-High Information (HI) animals in this experiment were presented with exteroceptive signals prior to impending shocks and following successful avoidance responses. Low Information (LI) did not receive these stimuli. It will be seen that HI animals achieved higher levels of Operant Efficiency than LI animals and thai they also reach asymptotic levels of Energy Efficiency more rapidly. Because the groups do not differ in their terminal rates of Energy Efficiency but the HI animals exhibit higher levels of Operant Efficiency, it follows that LI animals were committing more of the energy mobilized by the situation to performance of the ambulatory avoidance response. The results also suggest that information may be exchanged for work and energy expenditure.
504
Vol. 24. No. 5
Brener
•? 3 0 0 -
' 200 ..-cr'
Xf
FORCE
c- 1 0 0 -
0
1
2
3
4
5
0
T 1
\ 2
T 3
TIME S 50b— 40-
FORCE
30-
TIME 20-
10-
-1 1
r—r2 3
0 1 2 3-^ Hours in condi tioning SESSION 2 SESSION 1
Figure 8. Response measures used in beam-pressing experiment. Top Diagram: Peak Force = f,; Response Duration = t,-to; Impulse Strength = dF/dT = (f, -fo)/ (t|—to). Bottom Diagram: Integrated Force = area under curve {hatched area).
Figure 7. Changes in Operant and Energy Efficiency of avoidance behaviour in rats that were (HI) and were not (LI) provided with extra information about shock delivery and avoidance behaviour (from Sherwood. Brener, & Moncur. 1983).
series of experiments on beam pressing in rats.' In these experiments, subjects had to press a relatively rigid beam with a criterion force in order to earn food. The beam which consisted of an aluminum shaft to which strain gauges had been bonded was designed according to the specifications of Notterman and Mintz (1965). By continuously monitoring the force applied to the beam we were able to measure a full range of temporal and kinetic variables associated with the animals' task-related performance: this included responses that did not meet the criterion for reinforcement as well as those that did. The measures recorded are illustrated in Figure 8. Reinforcements which consisted of measured amounts of liquid food were delivered to a feeding cup on beam release in order to ensure that animals did not employ exteroceptive feedback associated with reinforcement delivery to control their motor performance. Figure 9 illustrates mean rates of energy expenditure, work, and reinforcement for 10 animals over the first 16 days during which they pressed the ofthe data from these experitnents is reported for the first time in this paper. They were conducted with Suzanne Mitchell, Sandra Readhead, and Mark Hutchinson.
Figure 9. Changes in reinforcement, work, and energy expenditure rates during acquisition of a beam-pressing response.
September, 1987
Behavioural Energetics
505
beam reliably." It will be seen that although work rates remained relatively constaTit, and reinforcement rates increased significantly, the rate of energy expenditure exhibited a systematic and reliable decrease. These observations, which resulted in significant decreases in work done and energy expended per reinforcement, conform to the Freeman and Sparrow models of efficiency changes during the course of learning. Improved performance was also marked by greater accuracy expressed by an early increase in IB 12 14 16 the relative frequency of responses that met the force criterion as well as by reductions in the variability DflTS of all the temporo-kinetic parameters illustrated in Figure 8 apart from Impulse Strength. Although Response Rate increased during the initial stages of acquisition, beam-presses tended to decrease in average magnitude over the course of the experiment. However, it was found that the kinetic variables. Peak Force and Impulse Strength, stabilized early in the acquisition process. On the other hand, as will be seen from Figure 10, Response Duration 1 declined systematically throughout the 16 days of observation. This was associated with a reliable de12 14 crease in response magnitude (Integrated Force) and DRTS hence in Work per Reinforcement. These results Figure 10. Changes in integrated force and response suggest that in this experimental context at least, more efficient performance, expressed by lower re- duration during 16 days of acquisition. inforcement costs, was based on modulation of Re- Duration. Although the net effect of these adjustsponse Duration. It is clear however that before this ments was to increase or maintain work output, feature of performance may be effectively regulated, they were accompanied by significant decreases in the response itself must have been formulated. overall energy expenditure. It is possible that unSeveral sets of data collected in this series of recorded energy-consuming responses were deexperiments suggest that the Force parameters (Im- creasing. Nevertheless, the results imply that motor pulse Strength and Peak Force) may be set on dif- learning is associated with the redistribution of enferent bases and stabilize earlier than Response Du- ergy resources. ration. This is illustrated in an experiment in which This example illustrates that work rates may inone group of 6 rats was required to press the beam crease or remain stable while the overall rate of for food with twice the force of another group of 6 energy expenditure falls. Other examples show that animals. The Impulse Strengths of the two groups work rates may be substantially augmented without proceeded along similar time courses to the same increasing rates of energy expenditure. For examasymptotes whereas the Response Durations of the ple, the data illustrated in Figure 12 were drawn Low Force Group decreased more than those of the from an experiment on 8 rats in which the force High Force Group (Figure 11) permitting each group requirement for reinforcement was augmented by to achieve Peak Forces that were appropriate to 4.12 grammes (the initial criterion) each week for their reinforcement criteria. This observation sup- 8 weeks. The results indicate that from the beginports the interpretation that Impulse Strength may ning to the end of this experiment the work rebe set on the basis of the mechanical properties of quirement went up by 8 times, the rate of work the effectors leaving the nervous system with only increased by almost 4 times, and yet the rate of a single degree of freedom to regulate—Response energy expenditure remained relatively stable.^ ••Animals were not provided wilh any explicit training but were simply placed in the experimental environment each day. Each beam-press that exceeded a criterion of 4,1 grammes Peak Force resulted in the delivery of reinforcement. The first day on which a subject earned 100 reinforcements in 60 min or less was designated Day 1 for that subject.
^Note that the energy expenditure is down to about 1,5 BMRs. This is after over 70 sessions of exposure to the experimental environment. Energy expenditure starts off at about 3 BMRs. rapidly decreases to about 2 BMRs, and then displays a gradual decrease to approximate the BMR. In behavioural experiments on freely-moving subjects it is usually between 1.5 and 2.5 BMRs.
506
Vol. 24. No. 5
Brener
IS
•a'
pr' /
]
2
3
4
S
B
7
6
9
10
11
12
13
DRTS
LOW FDRCE HIGH FORCE
9
IB 11 12 IS
DOTS
Figure 11. Changes in Impulse Strength, Response Duration, and Peak Force in animals required to press the beam with a Peak Force of 4.1 grammes (Low Force) and 8.2 grammes (High Force) for food. Note that the different response requirements do not influence impulse Strength but do affect Response Duration.
As in the previous example, in this case it might be that unrecorded activities were decreasing at a rate close to the rate at which recorded work was increasing and thereby compensating for the increase in task-relevant work. These unrecorded activities could include such processes as waste oxidation of free fatty acids (FFA) in which excess mobilized energy is burnt off by futile recycling between FFAs and triglyceride pools (Kjekhus, Ellekjaer, &. Rinde, 1980).'^ Speculative possibilities of this sort must be introduced in order to reconcile the dissociation of work and energy expenditure rates implied by the examples given. Without such possibilities the results imply that energy expenditure rates may be determined independently ofthe ''However, it could be that energy spared by lower levels of task-related work is employed to fuel task-irrelevant maintenance activities. Ethographic analyses of activity in an experiment by Sherwood, Brener, and Moncur (1983) showed that animals which ran less engaged in significantly more grooming activity than animals which ran more but burned the same amount of energy. Other data collected in an experiment with Moncur {Brener, 1986d) suggest that energy spared by task performance may be employed to capture information. That experiment suggested that animals will work for information as long as this does not decrease the energy efficiency of their taskrelated performance.
rate of work. However, as was mentioned, this idea is not uncommon in behavioural theory. The Effects of Variations in Outcome Uncertainty on the Magnitude of Responding
The experimental results described for the acquisition experiments agree with the observations reported by Notterman and Mintz (1965) that as training proceeds, there is a reliable tendency for the magnitude of task-related responses to recede toward the minimum levels required for task compliance. When these observations are considered in the light of the accompanying decreases in behavioural variability and increases in efficiency, they suggest an inverse relationship between the intensity of motor output and the degree of behavioural uncertainty: as effective motor responses are specified, redundant activities recede and effective behavioural variants come more-and-more to approximate the minimum requirements for reinforcement. Evidence of decreases in task-irrelevant activity was obtained in a recent experiment in which human subjects performed on a continuous choiceRT task for 4 sessions on each of which 720 trials were presented. They were required to switch off lights which were presented every 1.5 seconds by pressing one of four beams. We found (Figure 13)
September. 1987
4.1
Behavioural Energetics
8.2
12.3
16.4
20.5
24.6
28.7
32.8
2B.5 24.B 2B.7
32.8
FORCE REQUIREMENT ( g r * ) 1.6 T
1 . 7
••
l.B •
1.5 •
1.4
•!•
4.1
8.2
12.3
16.4
F O R C E R E Q U I R E M E N T (grtl
Figure 12. Changes in work and energy expenditure rates as a function of increasing force requirements for food. Note that work increases but not energy expenditure. PERCENT CORRECT
1 2 ! 4 SESSIONS
EHB (nicrovolts)
HE«T RflTE (bpn)
1 Z ! 4 SESSIONS
Figure 13. Changes in percentage of correct responses, EMG in unimplicated arm, and heart rate during learning of a multiple choice task in humans.
that as they learned the stimulus-response mapping and the latency and force criteria for task compliance, EMG decreased in the arm not implicated in pressing the keys as did heart rate and neck EMG (not illustrated). Thus as task proficiency increased, redundant striate muscular activity decreased. This finding conforms with tbe observations of a number of previous reports of decreases in static muscle tension during task mastery (e.g. Courts, 1942). Since isometric muscular contractions produce cardiac elevations through feedback and feedforward path-
507
ways (see Figure I6(C)), the decreases in irrelevant EMG may account for the fall in heart rate observed as task performance improved (Mitchell, Blomqvist. Lind, Saltin, & Shepherd, 1981). The concept of behavioural uncertainty has been explored by Germana (1972) who noted that autonomic activation is significantly augmented (i) in anticipation of an overt motor response, (ii) when subjects are required to change their responses to a stimulus, and (iii) that autonomic activation is directly related to the number of possible responses permitted by the prevailing requirements. Germana also makes reference to Bernstein's (1967) view that the process of motor learning involves reducing the number of degrees of freedom to be controlled, tbereby rendering the behavioural problem less complex and less uncertain. This analysis, which is compatible with that proposed here, suggests that experimental manipulations which increase task complexity and uncertainty will have the opposite effects to motor learning and will increase activation. In order to examine this hypothesis in relation to motor activation we decreased outcome uncertainty by lowering the probability with which a criterion response produced food from 1.0 to 0.25. According to the hypothesis that response magnitude is directly related to behavioural uncertainty, this manipulation should give rise to augmented responding. Tbe results of the experiment, which are illustrated in Figure 14, indicate that decreasing the rate of payoff led to profound increases in response magnitude (Integrated Force). Analysis of the response parameters revealed that Impulse Strength tended to decline slightly and that the response magnitude effect was achieved through the lengthening of Response Duration.^ It will be seen 'Although Response Rate showed a slight tendency to increase when the probability of Reinforcement was decreased from 1.0 to 0.25, the efFect was minimal and was not statistically significant. This conflicts with the widelyaccepted view that partial reinforcement leads to an increase in response rate. However, that empirical generalization is based on data collected using a conventional mieroswitch lever which operates as a binary force transducer. Since the response amplification process reported here would have pushed more responses over the binary criterion, the use of that behavioural transducer would have yielded an increased rate of response. This illustrates the point that quite different inferences about how the nervous system responds to changes in outcome probability are obtained by employing different methods of recording behaviour. With the conventional lever, it appears that the nervous system responds to increased outcome uncertainty by sending efference packets to the effectors at a higher rate whereas with the force transducer that we employed, it appears that the nervous system sends iai^er packets at approximately the same rate.
508
Vol. 24. No. 5
Brener 140 35 •
UB
30 •
80
25 •
SB
2a 2B
IE
IB
12
22
It
16
18
2B
22
18
20
22
DRTS
ORTS
2.6 T
6
2.4
2.2
10
12
14
IB DflYS
IB
2B
10
12
14 DBT5
Figure 14. Changes in petformance on changing the payoff probability from 1.0 (Days 10-13) to 0,25 (Days 14-22), Note substantial increases in work rate attributable to lengthening of responses wiih little change in energy expenditure.
that although work rates doubled, energy expend- havioural uncertainty, it will be observed that with iture did not change significantly and, indeed, continued exposure to the partial reinforcement showed a tendency to decrease. Thus as in the pre- contingencies, the response amplification process vious examples, the motor adjustment was made reflected by changes in these variables was attenby redistributing available energy resources and not uated. In other words, as the animals became faby mobilizing new resources. miliar with the lower payoff ratio, their response In order to understand how these mean per- magnitudes receded again toward minimal effective formance changes were effected, sequential analyses levels. were carried out on the response measures. They revealed a cumulative effect on response characDiscussion, Conclusions, and Speculations teristics of failure to achieve the goal object. Figure An attempt has been made here to develop an 15 shows the effects of consecutive failures to produce food on Impulse Strength, Response Dura- idea of how variations in metabolic rate are related tion, and Integrated Force. Separate curves are pre- to behavioural processes. Because of the imporsented for the first day on which the 0.25 payoff tance of striate muscular activity as a source of varwas applied, the fifth day, and the ninth day. The iation in energy expenditure, attention was focused tendency of Impulse Strength to decrease as a func- on the covariation of work and energy expenditure tion of consecutive unrewarded responses may be rates. Several sets of data indicate that these varidue to effector fatigue since the intervals separating ables are not always highly correlated. In general, consecutive failed responses tended to decrease. Re- energy expenditure tends to decline as a function spronse Duration, on the other hand, exhibited a of exposure to the experimental environment highly significant increase as a function of the num- whereas response magnitude decreases as a funcber of preceding failed responses. As in the previous tion of exposure to a fixed set of criteria for task experiments, the effects on this variable in partic- achievement. Thus metabolic rate may be said to ular may be held responsible for the changes in reflect the novelty of the environment whereas reIntegrated Force. In accordance with the hypothesis sponse magnitude reflects the novelty of the task that response magnitude is directly related to be- contingencies that prevail in the environment.
September, 1987
Behavioural Energetics
509
13S
115 •
2B
1
2
3
4
5
G
P
1
NUMBER OF PRECEDING FFIILURES
-D
SESSION 1 SESSION 5 SESSION 9
3
4
5
6
7a
NUMBER DF PRECEDING FfllLURES
7
a
2
•
6 S
•
4 3 2
NUMBER OF PRECEDING FRILURES
Figure 15. Changes in response charactetistics as a function of the number of preceding responses which had failed to earn food. The 0 point on the X-axis refers to responses whieh were preceded by a reinforced response.
Both of these processes may be viewed as expressions of behavioural uncertainty. Increases in metabolic rale may be related to environmental uncertainty and increases in motor intensity, to response uncertainty. Environmental uncertainty is concerned with the potential threats and benefits of the environment whereas response uncertainty is concerned with the motor activity required to meet these environmental challenges. In the analysis which follows, it is proposed that environmental uncertainty determines the rate at which energy will be mobilized to support adaptive motor activities. Response selection processes operate to reduce response uncertainty and thereby to restrict the motor channels through wbich mobilized energy will be expressed. Thus, it is suggested that when the organism first confronts a new situation, energy is mobilized by the release of catecholamines. Most of the energy is destined for the striate muscles but since appropriate responses have not yet been identified, the energy is distributed diffusely. This is expressed as highly variable motor performance which incorporates a pronounced static muscular component. One can speculate that tbe static muscular component of the initial motor adjustment functions to stabilize the body frame in preparation for unpre-
dictable challenges which may arise in the novel environment. The initially diffuse motor activation, indexed by high levels of variability in the various motor parameters recorded, may function to increase the chances of tbe organism encountering biologically-significant aspects of tbe environment and identifying effective coping responses. How environmental feedback may operate in the formulation of effective behavioural variants has been outlined elsewhere (Brener, 1986c). With regard to the level of the initial energy mobilization, it would seem that it has to be at least as great as the energy requirements of an effective coping response. If it is not, a successful behavioural variant cannot emerge. In this context I note our observation that animals which failed to acquire successful avoidance behaviour exhibited significantly lower energy expenditure increases on first encountering the shock contingencies than did successful animals {Brener et al., 1980). In freely-moving animals the initial rate of energy expenditure is about 3 BMRs with a range of 2-4 BMRS. Although I have not been concerned here with individual differences in tbe metabolic response, they may well have predictive value for other aspects of behaviour. Tbis matter has received attention from Gale (1986) who has suggested that tbe
510
Vol. 24, No. 5
Brener
coping styles exhibited by different personality types and associated with different psychological disorders may reflect individual variations in energetic processes. It seems reasonable to conjecture that the extent to which particular conditions mobilize energy in an individual provides a fundamental reflection of the behavioural significance of those conditions for that individual. In this context, it is also worth noting that hypertensives tend to have higher resting metabolic rates than normals (Lund-Johansen, 1980). It has been mentioned that during the initial stage of behavioural adaptation, behavioural uncertainty is greatest. This means simply that the subject doesn't know what to do about the situation. Before it can evolve an adequate coping strategy, the benefits and threats of the environment have to be identified and response-stimulus contingencies have to be perceived. Only when these things have happened can effective coping responses emerge and is it logically possible for the metabolic costs of an adequate behavioural adaptation to be estimated. We have found that as a function of practice, behavioural variability decreases and the specifications of effective responses come to approximate more-and-more closely, the minimum requirements for goal achievement. Furthermore, the rate of energy expenditure decreases. However, since recorded work rates tend not to decline during the acquisition process and indeed frequently increase, it is plausible that the rate of energy expenditure is not being driven exclusively by the immediate metabolic demands of the effectors. Several sets of data presented here support this interpretation. Substantial increases in work rate were achieved without significant alterations in energy expenditure. This occurred when work requirements were raised either by increasing the force requirements for reward or by decreasing the probability that a criterion response would produce a reward. Despite these elevations in work rate, energy expenditure remained stable or continued to decrease. This suggests that the rate of energy expenditure is determined by the novelty of the environment. Clearly, more precise characterizations of the parameters which determine the initial rate of energy expenditure and the rate of its subsequent decrease are required and this is a subject that we are currently investigating. One of the effects revealed in this research was the influence of outcome uncertainty on response magnitude. As subjects learned to cope with environmental requirements, response magnitude decreased. Conversely, when the probability of a successful outcome was decreased, response magnitude
increased. This process of response amplification is also reflected by electrocortical changes (Hink, Kohler, Deeeke, & Kornhuber, 1982; Kutas & Donchin, 1977). Empson's research reported at this meeting (1986) indicated that larger readiness potentials preceded larger movements and larger movements were provoked by a partial reinforcement schedule. Augmented eardiac responses have also been associated with conditions of uncertainty (e.g. Kahneman, Tursky, Shapiro, & Crider, 1969). They too might be expressions of motor amplification. The relationship between behaviour and energy expenditure depends critically upon cardiovascular functioning since active tissues must be perfused at rates that can support their levels of energy expenditure. When adaptive motor responses have been fully formulated, cardiac performance is highly correlated with metabolic rate and motor performance measures. This suggests that it is driven by feedback from the effectors implicated in meeting the task demands (Figure 16(A)). However, when the subject is faced with task demands that cannot be met with a preformulated response, cardiac performance appears to be elevated beyond metabolic requirements (Sherwood et al., 1986). This suggests that the motor planning processes which are required to resolve behavioural uncertainty may activate feedforward pathways by which descending influences from the motor control structures onto the medullary circuits controlling cardiovascular performance are effected (Figure 16(B)). Finally I return to the question of the relationship between rates of energy expenditure and striate muscular performance. Much of the data presented shows that it is not unusual for these two variables
HBTOR BRniH
*
NOTOR
»m
—}
CMDIOHRSCULUR CENTRES
CHDIOURSCyLM CENTRES
HEART m BLDOD VESSELS
NEMT AND BLOBD VESSELS
tiOTDR BRRIN CMlDIOURSCULRf! CENTRES
1 EIStCISINE HUSCLE
CA)
EtOtCISING HUSCLE
(B)
1 NEffIT m BLflBD ^ICSSELS EIERCISmE HUSCLE
CO
Figure 16. Serial and parallel models of sttiate muscular-cardiovascular interaction (after Mitchell, Schibye, Payne, & Saltin, 1981).
September, 1987
Behavioural Energetics
lo be uncorrelated." We should therefore consider a classical alternative to the serial model in which energy expenditure is a passive consequence of the rate of striate muscular activity. This alternative is shown in Figure 17. In this model, which is contained in the theories of Hull (1943). Hebb (1955) and Lorenz (1950) amongst others, the processes of response selection and energy mobilization are independent. The response selection processes determine the motor channels through which mobilized energy is expressed. How such response selection and energy mobilization processes interact remains unclear. An attack on this issue associated with a
"The observation that the frequency and intensily of striate muscular activity may change independently of alterations in the rate of energy expenditure indicates that the common practice of inferring rates of energy expenditure from rates of work in behavioural experiments is hazardous.
511
more comprehensive evaluation of the simpler serial model demands further research.
RELEftSINIJ STIMULUS
flCTION-SPECIFIC ENERGY
CUE
i
WDUSflL FUNCTIOM
INSTINCT
PERFWHRNCE
Figure 17. Parallel model of interaction between response selection and energy mobilization processes.
REFERENCES Astrand. P.O,. & Rodahl, K. (1977). Textbook of work bulation in freely-moving rats. Psychophvsiotogy. 17, physiology. New York: McGraw-Hill Book Company. 64-74, Banister. E.W.. & Brown. S.R, (1968). Exercise physiolBrener, J.. Phillips, K..C.. & Sherwood. A. (1983). Energy ogy. New York: Academic Press Inc, expenditure during response-dependent and responseBenedict. RG,. & Benedict. C.G, (1933). Menial effort in independent food delivery in rats, Psvchophvsiology. relation ro gaseous exchange, heart rale and mechanics 20. 384-392. of respiralion (Publication No, 446). Washington: Car- Cannon. W,B, (1929). Bodily changes in pain, hunger, fear negie Institution of Washington. and rage. New York: D. Appleton-Century Company, Bernstein, N, (1967). The coordination and regulation of Courts. F..\. (1942), Relations between muscular tension movements. New York: Pergamon Press. and performance. Psychological Bulletin, .i9, 347--367. Brener. J. (1986a). The circulation, behaviour and striate Duffy. E. (1951). The concept of energy mobilization. Psymuscular activity. The Behavioral and Brain Sciences, chological Review. 58. 30-40. 9, 296-297. Economos. A,C, (1982), On the origin of biological simBrener. J, (1986b), Factors influencing the covariation of ilarity. Journal of Theoretical Biology, 94. 25-60, heart rate and oxygen consumption. In P. Grossman. Empson. J.A.C. (1986). Response force, motivation and K.H.L, Janssen, & D. VaitI (Eds.), Cardiorespiratory the EEG readiness potential. Psvchophvsiology. 23. 433and cardiosomatic psychophysiology (pp. 173-190). 434. New York: Plenum Press. Engel, B.T. (1986). An essay on the circulation as behavBrener, J, (1986c), Operant reinforcement, feedback and iour. The Behavioural and Brain Sciences, 9. 285-295, the efficiency of learned motor control. In M.G.H. Freeman G,L. (1948), The energetics of human behaviour. Coles. E, Donchin, & S,W. Porges (Eds.). PsychophysIthaca: Cornell University Press, iology: Systems, processes and application.-; (pp. 309- Gale, A, (1986)- Arousal, control, energetics and values: 327). New York: The Guitford Press. An attempi at review and appraisal, in J, Strelau & Brener. J, (1986d). Behavioural efficiency: A biological H.J, Eysenck (Eds,). Personality dimensums and arousal link between informational and energetic processes. In (pp, 287-316). New York: Plenum Publishing CorGR.J, Hockey. ,\.W,K, Gaillard. & M.G.H, Coles poration. (Eds.). Energetics and human information processing Gellhorn. E, (1967), .iutonomic-somatic integrations. (pp. 113-122), Dordrecht: Martinus Nijhoff PublishMinneapolis: University of Minnesota Press. ers. Germana. J. (1972). Response uncertainty and autoBrener. J,, & Meakin. J. (1980), [The effects of anagram nomic-behavioral integration. .Annals of the New York solving on oxygen consumption and heart rate}. Un.U-ademy of Sciences, 193. 185-188. published study, Hebb. D.O. (1955). Drives and the CN.S. (conceptual Brener. J.. Phillips. K.C.. & Connally. S,R, (1980), Energy nervous system). Psychological Review, 62, 243-254. expenditure, heart rate and ambulation during shock Hess. E.W. (1954). Diencephalon, autonomie and periphavoidance conditioning of heart rate increases and ameral functions. New York: Grune and Stratton.
512
Brener
Vol. 24. No. 5
Hinde, R.A, (I960), Energy models of motivatioti. Pro- Mitchell. J,H., Schibye, B., Payne. FCIII.. & Saltin. B, ceedings of ihc Mceiing of the Sociely for E.xperimental (1981), Responses of arterial blood pressure to static B/(j/(j^?r (Symposium No. 14), 199-213. exercise in relation lo muscle mass, force development, and electromyographic activity. In J.H. MitchHink, R,F.. Kohler. H,. Deecke. L. & Kortihuber. H.H. ell. C.G. Blomqvist. ,A.R, Lind. B. Saltin, & J,T, Shep(1982), Risk-iakitig atid the human BcrcitschaftspoIcnlial, Elecu-oencephalography & Clinical Neurophys- herd (Eds.), Static (isometric) exercise: Cardiovascular responses and neural control mechanisms. Circulation iology. 53. 361-373. Research. 48lSupp\. I), 1-70 - 1-75. Hull, C,L, (1943). Principles of behavior. New York: .Appleton-Century-Crofts. Norberg. R,A. (1977), An ecological theory on foraging time and energetics and choice of optimal food search Kahnemati. D. Tursky. B., Shapiro. D.. & Crider. A. (1969). methods. Journal of Animal Ecology. 46. 511-529. Pupillary, heari rate and skin resistance changes during a mental task. Journal of Expcrinumia! Psychology. 79, Notterman, J,M,, & Mintz, D.E. (1965). Dynamics of re164-167. sponse. New York: John Wiley and Sons. Inc. Kjekhus, J,K., Ellekjaer, E.. & Rinde. P. (1980). The effect Obrist P.,A, (1976). The cardiovascular-behavioral interof free fatty acids on oxygen consumption in man: The action—As it appears today, Psvchophvsiologv. 13. 95free fatty acid hypothesis. Scandinavian Journal of 107. Clinical Laboratory Investigations. 40. 63-70, Rowell. L.B, (1974), Circulation to skeletal muscle. In T.C. Ruch & H.D, Patton (Eds,). Physiology and biophysics Kloiber. M. (1961), The lire of life: An introduction to (pp. 2(X)-214), Philadelphia: W,B. Saunders Company, animal energetics. New York: MacMillan, Kutas. M.. & Donchin. E. (1977). The efTect of handed- Sherwood. A., Brener, J,. & Moncur, D. (1983), Inforness. of responding hand, and of response force on the mation and stales of moior readiness: Their effects on contralaleral dominance of the readiness poiential. the covariation of heart rate and energy expenditure. Progress in Clinical Xeiifophysiology, I. 189-210. Psychophysiology. 20. 513-529, Langcr, A.W,. Obrist. P.A,. & McCubbin. J,A. (1979). Sherwood, A,, Allen, M.T., Obrist. PA,, & Langer, A,W, Hemodynamic and metabolic adjustments during ex(1986), Evaluation of beta-adrenergic influences on ercise and shock avoidance in dogs, American Journal cardiovascular and metabolic adjustments to physical of Physiology. 236. H225-H230. and psychological slrcss. Psychophysiology. 23. 89-104. Lewis, S.F., Taylor. W.F.. Graham. R.M,. Pettinger. W.A,. Sparrow, W,A. (1983). The efficiency of skilled performance, yozjr/ia/o/'Motor Sp/?fl\7o«r. 15. 237-261, Schutte, J,E.. & Blomqvist. C.G, (1983), Cardiovascular responses lo exercise as funetions of absolute and Stegemann. J, (1981). Exercise physiology. Chicago: Year relative workload. Journal of Applied Phvsiotogv. 54. Book Medical Publishers, Ine, 1314-1323. Stenberg. J.. Astrand. P,. Ekblom. B.. Royce J.. & Saltin, B, (1967). Hemodynamic response to work with difLorenz, K, (1950). The comparative method in studying ferent muscle groups, silting and supine. Journal of innate behaviour patterns. Proceedings of the .Meeting Applied Physiology. 22. 61-70, of the Society for Experimental Biologv (Symposium No. 4). 221-268. Turner J,R., & Carroll, D, (1985), Heart rate and oxygen consumption during mental arithmetic, a video game, Lund-Johansen, P, (1980). Haemodynamics in essential and graded exercise: Further evidence of metabolicalhypertension. Clinical Science. 59. 343s-345s. ly-cxaggerated cardiac adjustments? Psychophysiology, Martin, I,, & Levey, A,B. (1965), Effieiency of the con2!. 261-267. ditioned eyelid response. Science. 150. 781-783. Mitchell, J,H,, Biomqvist, C.G,, Lind, AR,. Saltin. B,, & Van den Berg, C.J, (1986), On the relation between energy Shepherd, J.T. (Eds.) (1981). Static (isometric) exertransformations in the brain and mental activities. In cise: Cardiovascular responses and neural control G.R,J, Hockey. A.W.K, Gailiard. & M.G.H. Coles, mechanisms. Circulation Research. •/5(Suppi. 1), 1-1 (Eds.), Energetics and human information processing 1-188. (pp. 131-135), Dordrecht: Martinus Nijhoff Publish-
Announcetnent Postdoctoral Research Associate The University of Texas Health Science Center at San Antonio has an immediately available position for a postdoctoral research associate to conduct psychophysiologieal studies of patients with muscle contraction headache. Studies will be conducted in the laboratory and in the natural environment using newly developed ambulatory EMG recording devices. Please send C.V. and examples of any scholarly work lo: John P. Hatch, Ph,D,, Research Director, Laboratory of Neuropsychophysiology. Department of Psychiatry, The University of Texas Health Science Center at San Antonio, 7703 Royd Curl Drive, San Antonio, TX 78284-7792.
