Brain Research Bulletin 66 (2005) 365–370

Cognitive and emotional functions of the teleost fish cerebellum

F. Rodrıguez∗, E. Duran, A. Gomez, F.M. Ocana, E.Alvarez, F. Jimenez-Moya,C. Broglio, C. Salas

Laboratory of Psychobiology, University of Sevilla, Campus Santiago Ram´on y Cajal,Avenida de San Francisco Javier, s/n. 41005 Sevilla, Spain

Available online 25 February 2005

Abstract

Increasing experimental and neuropsychological evidence indicates that the cerebellum of humans and other mammals, traditionallyassociated with motor control, is implicated in a variety of cognitive and emotional functions. For example, the cerebellum has been identifiedas an essential structure in different learning processes, ranging from simple forms of associative, sensory-motor learning and emotionalconditioning, to more complex, higher-order processes such as spatial cognition. Although neuroanatomical and neurophysiological dataindicate that the organization of the cerebellum is notably well conserved in vertebrates, little is actually known about the cerebellar contributionto processes besides the motor domain in non-mammals. In this work, we analyzed the involvement of the teleost fish cerebellum on classicalc ioning of as ysiologicalr Autonomice oldfish withc vable motord motor ore gly similart tion, havingb©

K n

vsoontmiFdi

cog-ut

otoromesh toition-e

basiceostver-ow-bel-

ofatial

0d

onditioning of motor and emotional responses and on spatial cognition. Cerebellum lesions in goldfish impair the classical conditimple eye-retraction response analogous to the eyeblink conditioning described in mammals. Single unit extracellular electrophecording and cytochrome oxidase histochemistry also reveal the involvement of the teleost fish cerebellum in classical conditioning.motional responses (e.g., heart rate classical conditioning) are also impaired by cerebellum lesions in goldfish. Furthermore, gerebellum lesions present a severe impairment in spatial cognition. In contrast, cerebellum lesions do not produce any obsereficit as indicated by the swimming activity or obstacle avoidance and do not interfere with the occurrence of unconditionedmotional responses. These data indicate that the functional involvement of the teleost cerebellum in learning and memory is strikin

o mammals and suggest that the cognitive and emotional functions of the cerebellum may have evolved early in vertebrate evolueen conserved along the phylogenetic history of the extant vertebrate groups.2005 Elsevier Inc. All rights reserved.

eywords:Cerebellum; Eyeblink classical conditioning; Emotional heart rate conditioning; Spatial memory; Teleost fish; Vertebrate brain evolutio

The cerebellum is one of the most variable parts of theertebrate brain in terms of shape and size. However, de-pite the notable macroanatomical variability, the cerebellumf every vertebrate species shows similar cytoarchitecturalrganization, neural circuits and neurophysiological mecha-isms[5,20]. Increasing neuropsychological and experimen-

al evidence indicate that the cerebellum of humans and otherammals, traditionally associated with motor control, is also

mplicated in a variety of cognitive and emotional functions.or example, it has been identified as an essential structure inifferent associative learning and memory processes, rang-

ng from simple forms of sensory-motor learning to more

∗ Corresponding author. Tel.: +34 95 455 77 46; fax: +34 95 455 77 52.E-mail address:fernanr@us.es (F. Rodrıguez).

complex, higher-order processes, for instance, spatialnition [15,23,30]. However, little is actually known abothe cerebellar contribution to processes besides the mdomain in non-mammalian vertebrates. Interestingly, searly lesion studies already related the cerebellum of filearning and memory processes, such as motor conding and avoidance learning[3,11]. In addition, besides somparticular features, the pattern of neuronal organization,circuitry and neurophysiological characteristics of the telfish cerebellum is similar to that of the other classes oftebrates[13,18]. Here, we summarize recent studies shing that in the teleost fish, like in mammals, the cerelum plays an important role in the classical conditioningsimple motor reflexes, in emotional learning and in spcognition.

361-9230/$ – see front matter © 2005 Elsevier Inc. All rights reserved.oi:10.1016/j.brainresbull.2004.11.026

366 F. Rodrıguez et al. / Brain Research Bulletin 66 (2005) 365–370

1. Eyeblink classical conditioning

The role of the cerebellum in classical conditioning (forexample, eyeblink classical conditioning) has been widelyinvestigated in mammals. In a typical eyeblink classical con-ditioning procedure, animals learn to express a conditionedresponse (CR) to a predictive or conditioned stimulus (CS)that is paired with a significant unconditioned stimulus (US).A considerable amount of evidence indicates that the essentialneural circuitry for acquisition and performance of this sim-ple, learned reflex resides in the cerebellum and related brainstem structures[17,30]. The contribution of the teleost fishcerebellum to classical conditioning has been investigatedrecently in a thorough and systematic study[8] (Fig. 1A).Cerebellum ablated and sham operated goldfish (Carassiusauratus) were trained in a delay classical conditioning proce-dure analogous to the eyeblink classical conditioning modelcommonly used in mammals. In this procedure, the CS wasa light (350 ms in duration), and the US consisted of a mild

perifacial shock (0.2–0.5 mA dc pulse, 0.15 ms in duration).In this paradigm, the CS onset precedes the US, but bothstimuli overlap in time and coterminate. The US evoked anunconditioned eye-retraction reflex or “eyeblink” (uncondi-tioned response, UR). Over the course of training the con-trol goldfish showed a progressive and significant increasein the percentage of conditioned responses to the CS pre-sentation, which were accurately timed to the onset of theUS. In the control goldfish, as in mammals, the percentage ofCRs increase with paired CS–US presentations and decreasewith CS alone (extinction) or unpaired CS–US presentations.The sensibility of the performance to these variations in thetraining conditions (i.e., in the CS–US relationships) indi-cates that learning was governed by associative rules andenables to disregard the possibility of pseudoconditioning bi-ases or other non-associative mechanisms. Interestingly, likein mammals, cerebellar lesions completely and permanentlyabolish eyeblink conditioning in goldfish. The cerebellumlesioned goldfish did not show significant changes in the per-

Faporttth

ig. 1. Goldfish cerebellum is involved in classical conditioning of simple motsevere impairment in delay eyeblink classical conditioning relative to the

ercentage of spontaneous eye-movement responses. The inset on the right sf the conditioned stimulus (CS) and the unconditioned stimulus (US) in the cate classical conditioning. Paired CS–US presentations produce an increaso the CS in the control goldfish. Goldfish with cerebellum lesions failed to ache bradycardia response to the CS during the habituation trials (autonomiche right shows electrocardiograms of representative sham and cerebellumeart rate deceleration response to the CS, the cerebellum lesioned animal

or reflexes and in emotional fear conditioning. (A) Cerebellum ablation producessham operated animals. In contrast, no significant changes were observed in thehows some illustrative examples of conditioned responses during paired presentationsontrol animals. (B) Cerebellum ablation also produces a severe deficit infear hearte in conditioned deceleration of the heart rate responses (conditionedbradycardia)quire the conditioned bradycardia response. However, no deficit was observed inorientation response to the CS) in the cerebellum lesioned goldfish. The inset onlesioned animals. Note that whereas the sham lesioned fish exhibits a conditioneddoes not. Asterisks (*) denote statistically significant differences.

F. Rodrıguez et al. / Brain Research Bulletin 66 (2005) 365–370 367

centage of CRs, independent of training conditions. In fact,the absence of increase in the number of CRs after 300 pairedCS–US training trials indicates a severe learning impairment.Furthermore, the deficit observed in the cerebellum lesionedgroup was selective to the CRs, as the percentage of URs, andspontaneous eye-movement responses was similar in bothgroups.

Additional evidence on the involvement of the teleostfish cerebellum in eyeblink classical conditioning wasprovided by extracellular single unit recording experiments.As in mammals, following CS–US paired conditioning, thePurkinje cells of the goldfish cerebellum exhibit learning-related changes in simple spike discharge frequency duringthe CS period that closely correlate with the onset of thebehavioral conditioned response[1]. In addition, in a recentexperiment, cytochrome oxidase (COX) histochemistrywas applied to study possible learning-related changes inthe cerebellar metabolic activity of goldfish[2]. Goldfishin the eyeblink classical conditioning group received 300CS–US paired trials during a 4-h session. Unpaired CS–USpresentation and untrained groups were used as controls.Following training, the animals were immediately perfusedand their brains processed for COX histochemistry. Opticaldensitometry analysis showed that the goldfish trained inthe paired condition presented an increase in the level ofCOX activity in the molecular and granular layers of thec andt COXa un-p ervedi weren lationo ningc

2

truc-t rtici-p s notr nse[ ion-i layc ingi re theC S iss timeg halics aged[

sso-c linkc bla-t fish[ link

classical conditioning in goldfish are critically dependent onthe cerebellum, irrespective of whether a delay or a trace pro-cedure is used[1,8,9]. Furthermore, lateral pallium lesions ingoldfish, like hippocampus lesions in mammals, selectivelyimpair eyeblink conditioning when a trace interval is intro-duced between the end of the CS and the onset of the US[9].Interestingly, the lateral pallium of teleost fish has been pro-posed as the homologue of the medial cortex or hippocampuson the basis of developmental, neuroanatomical and behav-ioral data[21,24,26]. Thus, these findings reveal that in teleostfish, like in mammals, the cerebellum plays an essential rolein classical conditioning independently of the temporal re-quirements of the procedure, whereas the hippocampal pal-lium is specifically involved in trace conditioning but not indelay conditioning.

3. Heart rate classical conditioning

Recently, a number of studies have suggested that the cere-bellum of mammals participates not only in the condition-ing of simple motor responses but also in emotional learning[7,29]. In a recent experiment, the involvement of the teleostfish cerebellum in emotional learning was studied by meansof analyzing the effects of cerebellar lesions on fear heart rateconditioning in goldfish[8] (Fig. 1B). In the control goldfish,p id in-c onses( l rel-a ringe le-s onse.I er int heartr ffectso fishs onse.S thec thouta se tot nedfi als,s ere-b sicalc

4

thatt strik-i -p gly,r m oft tialc oned

erebellum, relative to the fish in the unpaired conditiono the untrained animals. The absence of increase in thectivity of the cerebellar neurons in the fish trained in theaired CS–US condition suggests that the changes obs

n the cerebellum of the paired CS–US trained animalsot caused by unspecific factors such as sensory stimur stress, but, on the contrary, reflect specific learhanges.

. Delay versus trace classical conditioning

In mammals, the cerebellum and related brainstem sures mediate eyeblink classical conditioning, and the paation of brain structures above the level of the midbrain iequired for the conditioning of this simple motor respo22]. However, the independence of the eyeblink conditng from supracerebellar structures is valid only for deonditioning, not for trace conditioning. Trace conditionmposes additional task requirements, as in this proceduS and the US do not overlap; instead the end of the Ceparated from the onset of the US by a stimulus-freeap (trace interval). In these conditions, some telenceptructures, mainly the hippocampus, become eng12,19,28].

In teleost fish, like in mammals, the cerebellum and aiated brainstem circuitry seem to be sufficient for eyebonditioning in the delay paradigm. Thus, the complete aion of the forebrain spares this form of learning in gold9]. In addition, the acquisition and maintenance of eyeb

aired CS–US presentations consistently produce a raprease in the percentage of conditioned bradycardia respa deceleration of the heart rate during CS–US intervative to pre-CS baseline), which decreases quickly duxtinction training. In contrast, goldfish with cerebellumions failed to acquire the conditioned bradycardia respt is important to note that no deficit was observed neithhe autonomic orientation response to the CS nor in theate baseline in cerebellum ablated animals; that is, the ef the cerebellum lesions on the cardiac activity of goldeem to be selective to the conditioned bradycardia respimilarly, cerebellum lesions impair the acquisition ofonditioned bradycardia response in rats and rabbits, wiltering the heart rate baseline or the orientation respon

he CS[7,29]. Thus, the results from the cerebellum lesiosh are remarkably similar to those obtained in mammuggesting that the cerebellum of teleost fish, like the cellum of tetrapods, plays an essential role in the clasonditioning of emotional responses.

. Spatial learning

The results reviewed in the previous sections showhe mammalian and the teleostean cerebellum share ang functional similarity in classical conditioning of simle motor reflexes and in emotional learning. Interestinecent experimental results indicate that the cerebellueleost fish, like that of mammals, is also involved in spaognition. In one of these experiments, cerebellum lesi

368 F. Rodrıguez et al. / Brain Research Bulletin 66 (2005) 365–370

goldfish were tested in a spatial task analogous to the hole-board procedure commonly used in mammals[6] (Fig. 2A).Goldfish were trained to locate one baited feeder in a 25-feeder matrix, surrounded by an array of widely distributedvisual cues, which maintained stable spatial relationships rel-ative to the goal. Although the cerebellum lesioned goldfishshowed some improvement in their ability to locate the goal,they never reached the level of accuracy of the control andsham operated animals. Cerebellum ablated goldfish showeda stereotyped and inefficient exploratory behavior. Further-

more, the test trials indicate that cerebellum ablated goldfishsuffered a profound spatial cognition deficit, whereas shamoperated goldfish navigated directly to the goal despite the re-moval of any single visual cue, cerebellum lesioned goldfishfailed to readily reach the baited feeder when some particu-lar visual cues were excluded, indicating that they lack thecapacity to encode the goal location in a unique, map-likerepresentation of the environment. Similar results have beenobserved in mammals. For example, hemicerebellectomizedrats trained to locate a goal in a variety of spatial tasks, such as

Forcsoit

ig. 2. Goldfish cerebellum is involved in spatial cognition. (A) Goldfish withf the baited feeder within a regular 25-feeder matrix surrounded by an arrepresentative sham and cerebellum ablated animals from the start positioues. The small points show the location of the 25 feeders. (B) Goldfish wipatial and cue versions of a maze task. Diagrams represent both training cne remained open (goal) while a glass barrier blocked the other (dotted line

ndirectly, whereas in the cue procedure the goal was signaled directly. Theo the exit. Asterisks (*) denote significant statistical differences.

cerebellum ablation show a severe deficit in the accuracy to find the locationay of visual cues. Diagrams on the right show some examples of trajectories ofns (s1–s2) to the baited feeder (black circle); a–e, show the location of the visualth cerebellum ablation show a severe impairment in post-surgery training in theonditions. The box-maze had two start compartments and two exit doors but onlys). In the spatial procedure, the intramaze cue (stripped panels) signaled the goalcurved lines show the most appropriate trajectories from each start compartment

F. Rodrıguez et al. / Brain Research Bulletin 66 (2005) 365–370 369

the Morris water maze or the T-maze, never reach a high levelof performance and, in addition, show defective explorationpatterns and peripheral circling[10,14,23,25].

Additional evidence concerning the involvement of theteleost fish cerebellum in spatial behavior was provided by afurther experiment in which goldfish with cerebellum abla-tion, telencephalon lesions or sham operation, were trainedin a spatial or a cue learning task[6] (Fig. 2B). Typically,in these tasks, the deleterious effects of telencephalon abla-tion in teleost fish, and more specifically, of the hippocam-pal pallium lesions, are selective to the performance in thespatial task, sparing cue or orientation learning ([16,24,27];for reviews see:[4,26]). In contrast, the effects of the cere-bellum lesions on spatial performance are equally disruptiveregardless of training conditions; the impairment in the post-surgery performance is as severe in the fish trained in the cueprocedure as that observed in the fish trained in the spatialtask. The cerebellum lesioned fish show a dramatic decreasein the percentage of correct responses relative to the shamanimals and to pre-surgery level in both the spatial and thecue task. Indeed, the performance of these animals decaysabruptly to random levels, and does not show any significantimprovement during post-surgery training. Interestingly, theimpairment observed in the cerebellum lesioned goldfish isselective to spatial performance; no significant differenceswere observed in the distance traveled and swimming abil-i d thec thatw buts ancei

5

thatc hasb hobi-o enti so-c m-p ition.T d thet ty int mo-t n.

ionals fishc revi-o iono xam-p us tott narya of ap rte-

brate brain and behavior in which several, separated learningand memory systems dependent on distinct brain structuresand neural circuits may have evolved early in vertebrate evo-lution and been conserved through the phylogenetic historyof the extant vertebrate groups.

Acknowledgements

We wish to thank Mr. Gerardo Labrador for technical help.This work was supported by grants from the Spanish MCyTBFI 2001-3178 and AGL2000-0382-P4-05, and Junta de An-dalucıa CVI-242.

References

[1] E. Alvarez, A. Gomez, E. Duran, F.M. Ocana, F. Jimenez-Moya,C. Broglio, F. Rodrıguez, C. Salas, Brain substrates of “eyeblink”classical conditioning in goldfish, Acta Neurobiol. Exp. 63 (Suppl.)(2003) 62.

[2] E. Alvarez, A. Gomez, F. Rodrıguez, J.A. Gonzalez, H. Gonzalez-Pardo, J.L. Arias, C. Salas, Effects of classical conditioning on cy-tochrome oxidase activity in the cerebellum of goldfish, in: Inter-national Behavioral Neuroscience Meeting, vol. 11, Capri, 2002, p.49.

[3] L.R. Aronson, R. Herberman, Persistence of a conditioned responsee-

.brateionals.

Evo-

.arn-.15.rmis156.h.D.ber

.n-ts,

[ raintioralRes.

[ in

[ airsblink

[ , J.

[ ice,

[ es in

ty, obstacle avoidance, posture, between the control anerebellum lesioned animals. Thus, these results showhereas telencephalon lesions impair spatial memorypare cue learning, cerebellum ablation impairs performn both tasks.

. Concluding remarks

The traditional vision of the cerebellum as a structureontributes primarily to motor coordination and controleen challenged recently by broad evidence from psyclogy and clinical neuropsychology showing its involvem

n different cognitive and emotional functions, from asiative learning and emotional conditioning to more colex relational memory processes such as spatial cognhe results reviewed here show that the mammalian an

eleostean cerebellum share a striking functional similarihe classical conditioning of simple motor reflexes, in eional fear heart rate conditioning, and in spatial cognitio

Furthermore, the present results showing the functimilarity between the mammalian and the teleosteanerebellum in learning and cognition add strength to pus data indicating also a striking similarity in the functf other brain centers in teleost fish and amniotes, for ele, the pallial forebrain areas proposed as homologo

he hippocampus and the amygdala[4,26]. In addition, givenhat teleost fish and tetrapods share a common evolutioncestor, these similarities could reflect the conservationrimitive, common basic plan in the organization of the ve

in the cichlid fish,Tilapia macrocephala, after forebrain and cerbellar ablations, Anat. Rec. 138 (1960) 332.

[4] C. Broglio, A. Gomez, E. Duran, F.M. Ocana, F. Jimenez-Moya, FRodrıguez, C. Salas, Hallmarks of a common forebrain verteplan: specialized pallial areas for spatial, temporal and emotmemory in actinopterygian fish, Brain Res. Bull. (2005) in pres

[5] A.B. Butler, H. Hodos, Comparative Vertebrate Neuroanatomy:lution and Adaptation, Wiley-Liss, New York, 1996.

[6] E. Duran, A. Gomez, F.M. Ocana, E. Alvarez, C. Broglio, FJimenez-Moya, F. Rodrıguez, C. Salas, Cerebellum and spatial leing in teleost fish, FENS Forum Abstracts, vol. 2, 2004, p. A112

[7] B. Gherladucci, L. Sebastiani, Contribution of the cerebellar veto cardiovascular control, J. Auton. Nerv. Syst. 56 (1996) 149–

[8] A. Gomez, Neural bases of associative learning in goldfish, PDoctoral Thesis (unpublished), University of Sevilla, Decem2003.

[9] A. Gomez, E. Alvarez, E. Duran, F.M. Ocana, C. Broglio, FJimenez-Moya, C. Salas, F. Rodrıguez, Delay vs. trace conditioing following pallium ablation in goldfish, FENS Forum Abstracvol. 2, 2004, p. A042.10.

10] C.R. Goodlett, A.J. Nonneman, M.L. Valentino, J.R. West, Conston water maze spatial learning in rats: implications for behavstudies of brain damage and recovery of function, Behav. Brain28 (1988) 275–286.

11] H. Kaplan, L.R. Aronson, Function of forebrain and cerebellumlearning in the teleostTilapia heudelotii macrocephala, Bull. Am.Mus. Nat. Hist. 142 (1969) 141–208.

12] J.J. Kim, R.E. Clark, R.F. Thompson, Hippocampectomy impthe memory of recently, but not remotely, acquired trace eyeconditioned responses, Behav. Neurosci. 109 (1995) 195–203.

13] N. Kotchabhakdi, Functional circuitry of the goldfish cerebellumComp. Physiol. 112 (1976) 47–73.

14] R. Lalonde, M.I. Botez, Navigational deficits in weaver mutant mBrain Res. 398 (1986) 175–177.

15] R. Lalonde, M.I. Botez, The cerebellum and learning processanimals, Brain Res. Rev. 15 (1990) 325–332.

370 F. Rodrıguez et al. / Brain Research Bulletin 66 (2005) 365–370

[16] J.C. Lopez, C. Broglio, F. Rodrıguez, C. Thinus-Blanc, C. Salas,Reversal learning deficit in a spatial task but not in a cued oneafter telencephalic ablation in goldfish, Behav. Brain Res. 109 (2000)91–98.

[17] D.A. Mc Cormick, R.F. Thompson, Cerebellum: essential involve-ment in the classically conditioned eyelid response, Science 223(1984) 296–299.

[18] J. Meek, R. Nieuwenhuys, Holosteans and teleosts, in: R. Nieuwen-huys, H.J. ten Donkelaar, C. Nicholson (Eds.), The Central Ner-vous System of Vertebrates, Springer-Verlag, Berlin, 1998, pp. 759–937.

[19] J.R. Moyer, R.A. Deyo, J.F. Disterhoft, Hippocampectomy disruptstrace eyeblink conditioning in rabbits, Behav. Neurosci. 104 (1990)243–252.

[20] R. Nieuwenhuys, H.J. ten Donkelaar, C. Nicholson, The CentralNervous System of Vertebrates, Springer-Verlag, Berlin, 1998.

[21] R.G. Northcutt, The forebrain of gnathostomes: in search of a mor-photype, Brain Behav. Evol. 46 (1995) 275–318.

[22] D.A. Oakley, I.S. Russell, Subcortical storage of Pavlovian condi-tioning in the rabbit, Physiol. Behav. 18 (1977) 931–937.

[23] L. Petrosini, M.G. Leggio, M. Molinari, The cerebellum in spatialproblem solving: a co-start or a guest start? Prog Neurobiol. 56(1998) 191–210.

[24] F. Rodrıguez, J.C. Lopez, J.P. Vargas, Y. Gomez, C. Broglio, C.Salas, Conservation of spatial memory function in the pallial fore-brain of amniotes and ray-finned fishes, J. Neurosci. 22 (2002)2894–2903.

[25] L. Rondhi-Reig, N. Le Marec, J. Caston, J. Mariani, The role ofclimbing and parallel fibers inputs to cerebellar cortex in navigation,Behav. Brain Res. 132 (2002) 11–18.

[26] C. Salas, C. Broglio, F. Rodrıguez, Evolution of forebrain and spatialcognition in vertebrates: conservation across diversity, Brain Behav.Evol. 62 (2003) 72–82.

[27] C. Salas, C. Broglio, F. Rodrıguez, J.C. Lopez, M. Portavella, B.Torres, Telencephalic ablation in goldfish impairs performance in aspatial constancy problem but not in a cued one, Behav. Brain Res.79 (1996) 193–200.

[28] P.R. Solomon, E.R. Van der Schaaf, R.F. Thompson, D.J. Weisz,Hippocampus and trace conditioning of the rabbit’s classically con-ditioned nictitating membrane response, Behav. Neurosci. 100 (1986)729–744.

[29] W.F. Supple Jr., B.S. Kapp, The anterior cerebellar vermis: essentialinvolvement in classically conditioned bradycardia in the rabbit, J.Neurosci. 13 (1993) 3705–3711.

[30] R.F. Thompson, D.J. Krupa, Organization of memory traces in themammalian brain, Ann. Rev. Neurosci. 17 (1994) 519–549.