Author's personal copy A three-layered model of primate prefrontal cortex encodes identity and abstract categorical structure of behavioral sequences Xavier Hinaut ⇑ , Peter Ford Dominey Stem Cell and Brain Research Institute, INSERM U846, 18 Avenue du Doyen, Lepine, 69500 Bron, France Université de Lyon, Université Lyon I, 69003 Lyon, France article info Keywords: Echo state networks Reservoir computing Recurrent neural networks Laminar structure Sequence learning Primate Prefrontal cortex Striatum Model abstract Categoricalencodingiscrucialformasteringlargebodiesofrelatedsensory-motorexperiences,butwhat isitsneuralsubstrate?Inanefforttorespondtothisquestion,recentsingle-unitrecordingstudiesinthe macaquelateralprefrontalcortex(LPFC)havedemonstratedtwocharacteristicformsofneuralencoding ofthesequentialstructureoftheanimal’ssensory-motorexperience.Onepopulationofneuronsencodes the specific behavioral sequences. A second population of neurons encodes the sequence category (e.g. ABAB, AABB or AAAA) and does not differentiate sequences within the category (Shima, K., Isoda, M., Mushiake, H., Tanji, J., 2007. Categorization of behavioural sequences in the prefrontal cortex. Nature 445, 315–318.). Interestingly these neurons are intermingled in the lateral prefrontal cortex, and not topographicallysegregated.Thus,LPFCmayprovideaneurophysiologicalbasisforsensorimotorcatego- rization.Herewereportonaneuralnetworksimulationstudythatreproducesandexplainstheseresults. Wemodelacorticalcircuitcomposedofthreelayers(infragranular,granular,andsupragranular)of5 � 5 leakyintegratorneuronswithasigmoidaloutputfunction,andweexamine1000suchcircuitsrunningin parallel. Crucially the three layers are interconnected with recurrent connections, thus producing a dynamical system that is inherently sensitive to the spatiotemporal structure of the sequential inputs. Themodelispresentedwith11four-elementsequencesfollowingShimaetal.Weisolatedonesubpop- ulation of neurons each of whose activity predicts individual sequences, and a second population that predictscategoryindependentofthespecificsequence.Wearguethatarichlyinterconnectedcorticalcir- cuitiscapableofinternallygeneratinganeuralrepresentationofcategorymembership,thussignificantly extending the scope of recurrent network computation. In order to demonstrate that these representa- tionscanbeusedtocreateanexplicitcategorizationcapability,weintroducedanadditionalneuralstruc- turecorrespondingtothestriatum.Weshowedthatviacortico-striatalplasticity,neuronsinthestriatum could produce an explicit representation both of the identity of each sequence, and its category membership. � 2011 Elsevier Ltd. All rights reserved. 1. Introduction In1989thefirstreportofprefrontalneuronalactivityencoding thesequentialpropertiesofsensorimotorsequenceswasreported by Barone and Joseph (1989). They demonstrated that after the macaque had been shown a spatial sequence of three illuminated push-buttons,andwaspreparingtoreproducethesequence,neu- rons in the prefrontal cortex encoded both the spatial location of the targets, and most strikingly, their order or rank in the se- quence. Thus, the prefrontal neural population was considered to encode the entire action plan, necessary for the animal to repro- ducethevisualizedsequence.Sincethenanumberofstudieshave been performed, looking more closely at the sequence specific encoding of visuo-spatial behavioral components of sensorimotor sequences in the prefrontal cortex e.g. (Funahashi et al., 1997; Clower and Alexander, 1998). Recently, Shima and colleagues (2007) have considered that whenthesubjectisrequiredtoencodealargenumberofsensori- motor sequences, the increasing memory load would potentially requiretheanimaltoadoptacategorizationstrategy,sothesimilar sequences would be encoded as members of the same category. They employed a sequence reproduction task, where sequences were made up of the motor actions push, pull and turn, using a manipulable lever. The animal was required to be able to observe and reproduce 11 sequences from three categories: ‘‘paired’’ (AABB), ‘‘alternate’’ (ABAB) and ‘‘4-repeat’’ (AAAA), where A and Bwereeachsystematicallyreplacedbyoneofthethreeelements push, pull and turn. Four sequences were of the category AABB (e.g. Turn, Turn, Push, Push), 4 of the category ABAB and 3 of the category AAAA. In the experiment, the animal was presented one of these sequences under visual guidance five times, and then 0928-4257/$ - see front matter � 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.jphysparis.2011.07.010 ⇑ Correspondingauthor.Tel.:+33(0)472913482;fax:+33(0)472913461. E-mail address: xavier.hinaut@inserm.fr (X. Hinaut). JournalofPhysiology-Paris105(2011)16–24 Contents lists available at SciVerse ScienceDirect JournalofPhysiology-Paris journal homepage: www.elsevier.com/locate/jphysparis