Robertoux & Carlier, 2007

8/2/2019 Robertoux & Carlier, 2007

1/5

2007 EUROPEAN MOLECULAR BIOLOGY ORGANIZATION EMBO reports VOL 8 | SPECIAL ISSUE | 2007 S7

science & societyscience & society

From DNA to mindThe decline of causality as a general rule for living matter

Pierre L. Roubertoux & Michle Carlier

Genes are involved in psychol-ogy and behaviour, and in their

neuronal bases, in many spe-cies, including the geneticists favourites:human, mouse, Drosophila and the wormCaenorhabditis elegans. The number ofgenes that modulate behaviour in the mousealone is impressive: Table 1 summarizes thenumber of genes discovered on the basisof individual behavioural differences byexamining spontaneous mutations, targetedgenes and transgenic mice. A literatureanalysis of 23 journals revealed that, as oflate August 2006, 3,923 genes are involvedin mouse behaviour; when quantitative traitloci were included, the number increased tomore than 4,000. This is probably an under-estimate, as the analysis relied on such asmall collection of journals.

This number of genes gives rise to a para-dox: as of late August 2006, the number of

transgenic mice and mice with a targetedgene knockout identified in repositorieswas in the vicinity of 3,000: 2,500 wereidentified directly in repositories, and it wasestimated that another 500 were kept inlaboratories. As the mouse genome carriesabout 24,000 genes, a further 21,000 trans-genic or knockout lines are needed touncover the functions of all mouse genes.As most of the numbers in Table 1 wereestablished using transgenic or gene-target-ing technologies, we can therefore predictthe number of genes that are involved in

behavioural traitsafter fully exploring allknown genesby a simple calculation: thenumber of genes linked to behavioural traitsdivided by the number of targeted and trans-fected genes, multiplied by the estimatednumber of genes in the mouse, which is3,923/3,00024,000 = 31,384. Obviously,this is more than the total number of genes inthe mouse genome and is much larger than

the number of genes that could be expressedin the brain, regardless of the actual propor-tion of brain-expressed genes. A literatureanalysis of the same 23 journals reveals that3,900 additional genes have been recog-nized as modulating brain anatomy andfunction, with another 3,914 involved insensory abilities. If we include these genes inthe calculation, it would then predict thatsome 80,000 genes are involved in brainfunctioning and behaviour. The number istoo high: there is no room for this horde ofbrain-behaviour genes in the brain.

A comparison of the numerical ratios ofgenes to neurons in other species leads toa similar conclusion. The total number ofneurons in a species is an indicator of thecapability and behavioural complexity ofthe brain. However, the number of genes isnot predictive of the number of neurons andis therefore not predictive of the complexityof behaviour (Fig 1). Drosophila has fewer

genes than C. elegans, but more neurons anda larger behavioural repertoire. The mousehas fewer neurons and less behaviouralcomplexity than Homo sapiens, but approxi-mately the same number of genes. There isno correlation between the number of genesand the sophistication of behaviour or thenumber of neurons.

This mismatch between behaviouralcomplexity and the numbers of genesand neurons shows that the relationshipbetween genotype and phenotype is notlinear. The question then is what links DNA

There is no correlation betweenthe number of genes and thesophistication of behaviour orthe number of neurons

Table 1 | Number of reported genes and quantitative trait loci associated with behaviour in mice

Behaviour Gene Quantitative trait loci

Food and fluid consumption 334 17

Emotional reaction 341 112

Learning 306 38

Motor behaviour 1,542 43

Sleep and/or circadian activity 119

Social behaviour 223 11

Drug intake and related behaviour 254 33

Sensory and motor development 46 41

Seizures 301

Sensory behaviour 457

Total 3,923 295


2/5

EMBO reports VOL 8 | SPECIAL ISSUE | 2007 2007 EUROPEAN MOLECULAR BIOLOGY ORGANIZATIONS8

science & society specia l i s sue

at the most elementary level of organization

to behaviour at the highest level? An under-standing of these processes should cast lighton the intrinsic nature of the genebrainbehaviour link.

Thirty years ago, Changeux & Danchin(1976) suggested that the selectivestabilization of the synapse during

development could explain the relation-ship between the high number of special-ized neurons and the number of genes.Recent advances in molecular biologyoffer additional explanations. The conceptof the multifunctional gene proposes thatone gene might generate several behav-ioural or non-behavioural phenotypes(Roubertoux, 2004). We have found evi-dence for this by examining the number ofphenotypes associated with experimentalmutations. We reviewed 32 gene-targetingexperiments with mice conducted before1999 and compiled a list of the phenotypesreported for each of the 32 genes between1998 and 2006. Twenty-one genes areassociated with more than four behav-ioural phenotypes each. One example isthe -CamkII gene, which modulates dif-

ferences in learning, anxiety, aggressionand sensitivity to pain, and which affectslong-term potentiation in the hippocampusas well as several non-behavioural pheno-types. Eleven other genes associated withmore than two behavioural phenotypeseach are also linked with several muscular,skeletal, immunological, digestive and res-piratory phenotypes. Multifunctional geneshave also been found in other species: theclk-1 gene in C. elegans affects egg-laying,pharyngeal pumping, crawling, swimmingand defaecation. A single mutation in the

EPM2A human gene induces seizures, dys-

lexia and learning defects (Ganesh et al,2002). The multifunctionality of the geneis thus a general property throughout thegenome and means that a vast phenotypicdiversity can be produced from a smallnumber of genes.

Multifunctionality extends the concept ofpleiotropy, whereby a polypeptide has differ-ent effects in different cells. It would explainwhy the same allele induces one effect ina liver cell and another in a heart cell. Thebrain has widely diversified functions and itsparts are as different from each other as theyare from other organs. The same allelic formof a gene thus produces different effects inbrain structure A than in brain structure Bbecause the differences between A and Bcan be as great as the differences between aheart cell and a liver cell.

This was illustrated in mice with anull allele for the dopamine hydroxy-lase gene, which produces a dopaminedeficiency throughout the body. A totallack of dopamine induces a variety ofdisorders, including feeding deficits,sucrose avoidance and poor interest innest building. Normal behaviour pat-

terns can be restored by locally injectingdopamine into the regions of the caudateputamen and nucleus accumbens. Whendopamine is injected into the lateral cau-date putamen, feeding is restored; whenit is injected into the nucleus accumbens,only sucrose preference is restored; andwhen it is injected into the lateral or thecentral caudate putamen, nest building isrestored (Szczypka et al, 2001). The prod-uct of the gene therefore has various func-tions depending on the brain regions inwhich the gene is expressed.

Alternative splicing has been suggested

as a mechanism behind this multifunction-ality (Roubertoux, 2004; Ule et al, 2006).Through alternative splicing, the DrosophilaDown syndrome cell adhesion molecule(Dscam) gene, which encodes an axonguidance receptor, can express 38,016 dif-ferent messenger RNAs and thus 38,016possible functions (Schmucker et al, 2000).This is three times the size of the Drosophilagenome, which has some 13,000 genes.Other genetic events contribute to themultifunctionality of genes either aloneor in interaction with alternative splicing.Interactions between the allelic forms of agene or of several genes and genetic cas-cades increase the number of genotypesthat subsequently foster the emergence ofmultiple phenotypes. Intra-locus interac-tion, as observed in genomic imprinting,induces different phenotypes dependingon whether the imprinted allele is from thefather or the mother.

The interaction between alleles ofdifferent genes, called epistasis,produces a larger set of possible

genotypes; for example, two genes A and

B, each with two alleles, can generate ninegenotypes (AABB, AABb, AAbb, AaBB,AaBb, Aabb, aaBB, aaBb, aabb). When asingle gene is either suppressed or presentas multiple copies, it can affect the expres-sion of several hundred other genes, thusgiving rise to numerous phenotypes. Forexample, two mitochondrial genes codingfor proteins show non-synonymous differ-ences between two mice strains; the cross-transfer of mitochondrial DNA betweenthe strains modulates cognitive perform-ance in the resulting congenic mice

A B

0

100

50

15,000 30,000

20,000

5 106

302

109

Numbe

rofneurons

Num

berof

familiesofionchannels

Drosophila

Drosophila

Caenorhabditiselegans

Caenorhabditiselegans

Musmusculu

sdomesticus

Homosapiens

Homo

sapiens

Musmusculusdomesticus

Numberof

receptor-coupledGproteins

Number of genes

1,100

700

100

0

Drosophila

C.elegans

Musmusculu

sdomesticus

Hom

osapiens

C

Fig 1 | Interspecies relationships showing the ratio of the number of genes to the (A) number of neurons, (B) ion channel families and (C) categories of G proteins

(adapted from Roubertoux, 2004).


3/5


science & societyspecia l i s sue

(Roubertoux et al, 2003). The modulationseems to be caused not only directly by themitochondrial alleles, but also indirectlyby interactions with nuclear genes. Of the34 behavioural measurements made, 16showed a direct effect from mitochondrialDNA, and 15 showed interactions withnuclear genes.

The role of epistasis in generating multi-functionality could also explain the differ-ences between humans and chimpanzees.About 1.34% of genes differ between theseclosely related species, and about 100 ofthese genes are expressed in the brain. Couldthese 100 genes encompass the genes of thehuman species, including those necessaryto discover the existence of the double helixor to compose Tristan and Isolde? It is nowknown that these 100 genes interact withgenes common to both humans and chim-panzees and that they modify the expressionof 35% of the chimpanzee genesapproxi-mately 8,400most of which are expressedin the brain (Hacia, 2001).

Cascade is a specific form of epistasis:the function of gene requires the productof gene , and the function of both and requires the product of gene (Fig 2).Cascades frequently affect genes involved indevelopment. The paired box 6 (PAX6/Pax6)gene is regulated by five genes and itself reg-ulates more than 30 genes, thus indirectlycontributing to several phenotypes. Thesame is true for the steroid sulphatase gene(Sts), which we identified as contributing tospontaneous aggression against a male ofthe same species (Roubertoux et al, 2005).

We realized that the Sts gene was the initia-tor of a cascade of biological events, ratherthan being directly responsible for aggres-sive behaviour. Steroid sulphatase catalysesthe hydrolysis of 3--hydroxysteroid sul-phates. The sulphated steroids modulate typeA -aminobutyric acid receptors (GABA

A)

and N-methyl-D-aspartate A receptors(NMDA

A), which are known for their role

in aggression (Maxson, 1996). Dependingon whether these steroids are sulphated,GABA

Aand NMDA

Acontribute to a wide

range of neuronal mechanisms involved inneuronal plasticity, learning, cognition andmotor behaviour.

Regardless of how genes interact, it is dif-

ficult to identify the one gene that triggersa chain of biochemical and cellular eventsthat eventually lead to a phenotype. It istherefore absurd to ascribe a certain pheno-type to a single gene in cases in which thephenotype does not display the main effectof the gene in question. Instead, a geneticnetwork, not a single gene, produces thephenotype in almost all brain-related pheno-types. We should thus consider the genomeas a site where genes interact in constantdialogue and exchange, and not as a staticalignment of functional units.

The concept of genotypeenviron-ment interactions is often used toexplain phenotypic variationand

even more often misused. It is the trumpcard for zealous environmentalists and atool for unrepentant advocates of heredity,and provides a safe haven for the virtuousdefendants of ecumenism. Genotypeenvi-ronment interactions cannot operate in onedirection only, in which the environmentaffects the gene. Instead, the effect of a genedepends on the environment in which theorganism develops and lives, but the sameenvironment does not have the same effectson different genes. Studies investigating thecombined use of ovary transplantation plusadoption in mice showed that certain genestend to accelerate development, but theeffect was modulated by a complex equi-librium between components in the mater-nal milk and the quality of care, such as thefrequency of licking and lactating episodes,and nesting characteristics.

This does not necessarily mean that one

environment is always better than another. Arich milk composition and frequent care canhave beneficial consequences for certaingenotypes, but will have null or negativeeffects on others (Carlier et al, 1999). In fact,most of the genes involved in behaviouraldifferences are susceptibility genes: theenvironment increases the effect of the geneand the diversity of environments mightincrease the diversity of phenotypes pro-duced by that gene. Overexpression or under-expression of genes in different environmentalconditions is the most probable mechanism

A B

A

B

C

BRAIN

a

b

c

BEHAVIOURGENOME

A

B

C

BRAIN

a

b

c

BEHAVIOURGENOME

123

1

23

12

3

Fig 2 | The relationship between genes, the brain and behaviour. (A) An obsolete view assuming isomorphism between the three levels. (B) A more recent view

based on nonlinear relationships. Each gene, and has several allelic forms that can undergo alternative splicing as shown by the small arrows; genetic cascades

are the vertical arrows. The resulting brain modules, A, B and Cconsisting of one neuron or a set of neuronsinteract as indicated by the vertical arrows. Each

module, either alone or in combination with other brain modules, can produce differences in one or more behaviour patterns. The modification in behaviour

regulates brain functioning and modulates gene expression, as illustrated by the lower curved arrows.

The product of the gene thereforehas various functions dependingon the brain regions in which thegene is expressed


4/5

EMBO reports VOL 8 | SPECIAL ISSUE | 2007 2007 EUROPEAN MOLECULAR BIOLOGY ORGANIZATIONS1 0

science & society specia l i s sue

of genotypeenvironment interactions. Modi-fications induced in the brain during learningindicate how the brain is able to integratenew rules and develop new strategies by trig-gering changes in neuronal connectivity andlong-term changes in gene expression.

The onset of a certain phenotyperequires proteins, which are the finalstage in a series of transitional steps

from transcription to translation. Each stepis an integrative process, whereby the gen-eration of information at one level relies oninformation provided by previous elemen-tary steps. The price for this generation ofinformation is weaker causality of the linkfrom DNA to protein in a given cell, whichhas several immediate consequences.

First, most of the genes involved in inte-

grative functions are susceptibility genes. Theallelic form associated with the integratedphenotype does not determine the pheno-type, but defines a lesser or greater probabilityfor expressing the phenotype. There are someexceptionsa person carrying one allele of aspecific gene with more than 40 CAG repeatswill develop Huntington disease; the numberof repeats does not, however, predict the ageof onset, the severity of the disorder or therate of neurodegeneration.

Second, most phenotypes require sev-eral alleles acting cumulatively or by inter-action. The probability of receiving from theparents all alleles necessary to develop agiven phenotype is lower than the probabil-ity of receiving alleles in the case of mono-genic traits. Genetics does not necessarilymean heredity.

Third, if causality were linear, it shouldbe possible to deduce the phenotype fromthe genotype and to infer the genotype fromthe phenotype. This is not the case for mostphenotypes. A mouse with an NZB/BlNJ Stsallele, for example, has a high probabilityof attacking males from the same species(Roubertoux et al, 2005). In turn, allelic var-

iants for more than 50 genes can be inferredfrom the fact that a male mouse attacksanother male mouse (Maxson, 1996). Ingeneral, the knowledge of a persons geneticmake-up might be used to predict the prob-ability of a certain phenotype, but in mostcases it is impossible to make a posteri-oricalculations of the genetic variants onthe basis of the phenotype. In humans, forexample, it would not be possible to inferthe mutation associated with a given intel-lectual disability phenotype if more than1,370 loci are known to be involved.

Fourth, genetic analysis consists of cor-relations between genetic differences andphenotypic differences. We cannot finda gene encoding a given phenotype, butwe can detect alleles linked to individualphenotypic differences. Genetic analysiscannot deal with invariants; it operateswith variations. We can discover a group ofgenes involved in learning and memory, buttogether, these genes do not constitute thegenetics of memory in a given species.

The same absence of isomorphismobserved for genotype and phenotype isalso found for the relationship between thebrain and behaviour. The elementary com-ponents of a neuron do not predict behav-ioural complexity (Fig 1). The number ofion channel families in C. elegans is greaterthan in Drosophila, but Drosophila is capa-ble of more complex behaviour. C. eleganshas a greater variety of G proteins thanDrosophila, but only the latter is capable ofdiscriminative learning.

It is the higher number of neurons inmammalian species, and particularly inhumans, that leads to greater complexity

and thus makes a difference in behaviour.The large number of possible connections,and therefore the large number of synapses,means there are more possibilities for neuro-nal integration, the impact of alternativesplicing and more frequent cascades.

For several decades, scientists acceptedHenry Dales principle of one neuronreleases one transmitter as they followedthe one gene codes for one protein rule,

but, with time, both were questioned. Wenow know that one neuron can influencemore than one transmitter system, andthat one neurotransmitter can modulatethe release or reuptake of another neuro-transmitter. Modern neurochemistry, byfocusing on interactions between neuro-transmission systems, offers an analogywith epistasis and cascades, which leads toa similar dilution of causality.

The quest to find a relationship betweena given transmitter and a given feelingor behaviour has now been abandoned,

because the effect of each neurotransmittercan vary depending on where it is released.In two separate cases, for example, the sameinterneuron can induce different behav-ioural responses in the mollusc Tritoniadiomedea: rhythmic swimming, which usesthe muscles; or non-rhythmic crawling,which uses the ciliar system. The versatil-ity or multifunctionality of the interneurondepends on the neuronal context; whetherthe interneuron triggers swimming or crawl-ing depends on the activity of its adjacentneurons (Popescu & Frost, 2002). Similarly,the concept of a reflex supported thehypothesis of ultra-specialized, hard-wired neuronal networks. For example, thestretching reflex was seen as the prototypeof an autonomous, specialized function, butit now seems that the sensory-motor units

provide the appropriate response and arethe elementary components of the stretch-ing response, rather than the motor neurons(Clarac et al, 2000).

The concept of multifunctional genesgoes together with the concept ofmultifunctional neurons or neuronal

networks. When it is combined with cas-cades, or with interactions between neuro-transmitters, the possibility of predictingbehavioural differences on the basis of braincharacteristics is undermined. Immediatelyafter the sequence of the human genomewas published, scientists thought that therewould be a simple relationship between thegenome, the nervous system and behaviour.However, this idea quickly became obsolete(Fig 2A) as it was based on linear relation-ships between genes, the brain and behav-iour. A more realistic view holds that eachgene, , and , has several allelic forms(Fig 2B); each allelic form can undergoalternative splicing; and the genes act incascades. A gene might act either alone orcumulatively, interacting with other geneson one or several brain phenotypes. The

resulting brain modules again interact andeach module, either individually or in com-bination with other brain modules, can pro-duce differences in one or more behaviourpatterns. The modification in behaviour thenregulates brain functioning and modulatesgene expression.

Genes are obviously involved in gen-erating behaviour, but genetic deter-minism is rare. We can only predict theprobability of a given phenotype on thebasis of a gene or, more usually, a networkof genes. Accordingly, we must change our

It is [] absurd to ascribe acertain phenotype to a single genein cases where the phenotypedoes not display the main effectof the gene in question


5/5


science & societyspecia l i s sue

terminology: no gene codes for a behaviouraltrait; at the very most, we can say that a geneis involved in a behavioural trait, or that aphenotype has genetic correlates. In thesame vein, the terms genetics of behaviouror behavioural genetics seem inappropriateas they infer strict determinism; indeed, theyare absurd as behaviour carries no genes,for it is the individual who carries the genes.Behaviour-genetic analysis (Hirsch, 1963)seems to be a more appropriate term: insteadof focusing on relationships between genesand behaviour, it uses individual differencesin behavioural phenotypes to discover allelicforms and their function.

What does natural selection then meanin the context of multifunctional genes whengenephenotype causality is tenuous? It ismuch more difficult to understand the effi-

ciency of selection when the gene on whichselection is supposed to act has several func-tions, some contributing to positive adapta-tion and others to negative adaptation.

The nature of causal relationshipsbetween genes and behaviour alsoraises concerns wherever society uses

scientific findings to allay ancestral fears,to encourage us to move into a brave newworld, or to use the authority and aura ofscience to conceal ulterior motives. A reportby the European Commission discussed thepossibility of detecting susceptibility genesfor occupational diseases (EGE, 2000);carriers of a gene that increases the risk ofdeveloping such a disease might then be dis-couraged from continuing in their positions. Job applicants might be rejected for thesame reason. The emphasis on being able todetect genetic factors and the disregard forweak causality would mean that employerscould ignore the need to find a proper solu-tion, such as abandoning the practices andtechniques that cause occupational diseasesin the first place.

The decline in genephenotype causal-

ity and the emergence of multifunctionalitymust also raise questions in medical appli-cations, including medically assisted repro-duction, gene therapy, therapeutic cloningand cell cloning. These techniques have notyet taken into account all the possible rami-fications, such as epistatic effects, betweenthe corrected gene and the host genome.This is particularly true for therapeutic clon-ing, for which the role of mitochondrial DNAand its modulating effects on nuclear geneshave been underestimated. Mitochondrial

DNAwhich, in most cases, is of unknownoriginhas been shown to have unexpectedeffects (Roubertoux et al, 2003). Cloningmight therefore generate phenotypes thatdiffer from those expected from the nucleargenotype alone. There is a clear need forbasic research on the epistatic events thatare generated by mitochondrial DNA beforewe can safely move on to generate tissue fortransplantation by therapeutic cloning.

Other medical issues, such as geneticcounselling, prenatal diagnosis and embryoselection, are also affected if the strict rela-tionship between gene and phenotype dis-solves. Direct-to-consumer genetic tests canscreen for monogenic disorders, susceptibil-ity genes and polygenic disorders. However,the causality between susceptibility genes,the components of polygenes and pheno-

type is weak. Most disorders are caused byseveral genes, and these genes vary from onepopulation to another. How is it possible tomake any predictions? Similarly, rejectingan embryo with undesirable characteristicson the basis of a prenatal diagnostic test isnot a simple decision because it does notdepend solely on negative parameters, aswas previously thought with the one gene,one protein, one function perspective. Anindividual is not determined by one sin-gle deleterious gene but through personaldevelopment, and living with and reactingto the genetic disease. If direct-to-consumergenetic testing and prenatal diagnostic testshad been available centuries ago, and ifpeople carrying genes associated with epi-lepsy had been eliminated, the world wouldnever have known Leonardo da Vinci, LeoTolstoy, Gustave Flaubert, Graham Greeneor Vincent van Gogh.

ACKNOWLEDGEMENTSThis article is dedicated to Professor Jean-MarcMonteil with esteem and friendship.

REFERENCESCarlier M, Roubertoux PL, Wahlsten D (1999)

Maternal effects in behavior genetic analysis. InJones BC, Mormde P (eds) NeurobehavioralGenetics: Methods and Applications, pp 187197.Boca Raton, FL, USA: CRC

Changeux JP, Danchin A (1976) Selectivestabilisation of developing synapses as amechanism for the specification of neuronalnetworks. Nature264: 705712

Clarac F, Cattaert D, Le Ray D (2000) Centralcontrol components of a simple stretch reflex.Trends Neurosci23:199208

EGE (2000) Genetic testing in the workplace.Proceedings of the round table debate held at theBrochette Center, Brussels. Brussels, Belgium:

European Group on Ethics in Science and NewTechnologies, European Commission

Ganesh S et al(2002) Genotypephenotypecorrelations for EPM2A mutations in Laforasprogressive myoclonus epilepsy: exon 1mutations associate with an early-onsetcognitive deficit subphenotype. Hum Mol Genet

11: 12631271Hacia JG (2001) Genome of the apes. Trends

Genet17: 637645Hirsch J (1963) Behavior genetics and individuality

understood. Science142: 14361442Maxson SC (1996) Searching for candidate genes

with effects on an agonistic behavior, offense, inmice. Behav Genet 26: 471476

Popescu IR, Frost WN (2002) Highly dissimilarbehaviors mediated by a multifunctionalnetwork in the marine mollusk Tritoniadiomedea.J Neurosci22: 19851993

Roubertoux PL (2004) Existe-t-il des Gnes duComportement? Paris, France: Odile Jacob

Roubertoux PL et al(2003) Mitochondrial DNAmodifies cognition in interaction with the

nuclear genome and age in mice. Nat Genet35:6569Roubertoux PL, Guillot PV, Mortaud S, Pratte

M, Jamon M, Cohen-Salmon C, Tordjman S(2005) Attack behaviors in mice: from factorialstructure to quantitative trait loci mapping. Eur JPharmacol526: 172185

Schmucker D, Clemens JC, Shu H, Worby CA,Xiao J, Muda M, Dixon JE, Zipursky SL (2000)DrosophilaDscam is an axon guidance receptorexhibiting extraordinary molecular diversity. Cell101: 671684

Szczypka MS, Kwok K, Brot MD, Marck BT,Matsumoto AM, Donahue BA, Palmiter RD(2001) Dopamine production in the caudateputamen restores feeding in dopamine-deficientmice. Neuron30: 819828

Ule J, Stefani G, Mele A, Ruggiu M, Wang X, TaneriB, Gaasterland T, Blencowe BJ, Darnell RB(2006) An RNA map predicting Nova-dependentsplicing regulation. Nature444: 580586

Pierre L. Roubertoux is in the FunctionalGenomics Group, CNRS Aix-Marseille Universityin Marseille, France.Michle Carlier is at the University Institute ofFrance and Laboratory of Cognitive Psychology,CNRS, Aix-Marseille University, France.E-mail: [email protected]

doi:10.1038/sj.embor.7400991

Robertoux & Carlier, 2007

Documents

Transcript of Robertoux & Carlier, 2007