Sit-et-al-2009Complex Dynamics of V1 Population Responses Explained by a Simple Gain-Control ModelIntroductionResultsPopulation Responses to a Gabor Stimulus in V1Peak Responses: Spatial Distribution and ContrastOverview of the Temporal Response Properties atnbspDifferent LocationsProperties of the Rising EdgesProperties of the Falling EdgesPlausible Families of ModelsStatic Nonlinearity ModelsLateral Propagation ModelsNormalization ModelsPopulation Gain-Control ModelProcessing in the Model UnitsResponse Transformation between StagesGeneral Behavior of the ModelEffects of Normalization Pool SizeSimulation of VSDI ResponsesRelative Normalization Strengths in the Different StagesDiscussionRelationship between the Responses of a Single Neuron and a Neural PopulationPossible Implementation of Divisive Population Gain-ControlTraveling Wave of Neural ResponsesConclusionExperimental ProceduresBehavioral Task and Visual StimulusAnalysis of Imaging DataModel DefinitionProcessing in Each Model UnitResponse TransformationSimulation of the ModelSupplemental DataAcknowledgmentsReferencesSit-et-al-2009-supplementSit-et-al-2009-commentaryThe V1 Population Gains NormalizationReferencesNeuronArticleComplex Dynamics of V1 Population ResponsesExplained by a Simple Gain-Control ModelYiu Fai Sit,1Yuzhi Chen,2Wilson S. Geisler,2Risto Miikkulainen,1and Eyal Seidemann2,*1Department of Computer Sciences2Department of Psychology and Center for Perceptual SystemsThe University of Texas at Austin, 1 University Station, A8000, Austin, TX 78712, USA*Correspondence: [email protected] 10.1016/j.neuron.2009.08.041SUMMARYTo understand sensory encoding and decoding, it isessential to characterize the dynamics of populationresponses in sensory cortical areas. Using voltage-sensitive dye imaging in awake, fixating monkeys,we obtained complete quantitative measurementsof the spatiotemporal dynamics of V1 responsesover the entire region activated by small, brieflypresented stimuli. The responses exhibit severalcomplex properties: they begin to rise approximatelysimultaneously over the entire active region, butreach their peak more rapidly at the center. However,at stimulus offset the responses fall simultaneouslyand at the same rate at all locations. Althoughresponse onset depends on stimulus contrast, boththe peak spatial profile and the offset dynamics areindependent of contrast. We show that these resultsare consistent with a simple population gain-controlmodel that genera lizes earlier single-neuron contrastgain-control models. This model provides valuableinsight and is likely to be applicable to other brainareas.INTRODUCTIONSmall visual stimuli elicit neural responses that are distributedover a large area in the primate primary visual cortex (V1; e.g.,Hubel and Wiesel, 1974; Grinvald et al., 1994), suggesting thateven small stimuli are encoded by a large population of neuronsin V1. Furthermore, electrophysiological studies in behavingprimates suggest that perception is mediated by a populationof neurons rather than by single neurons (Parker and Newsome,1998; Purushothaman and Bradley, 2005 ). Thus, to understandthe encoding and decoding of visual stimuli in the cortex, itis important to characterize the properties of V1 populationresponses.One approach is to estimate population responses from singleneuron responses. Single unit recordings in V1 have revealeda number of fundamental properties that ought to contribute tothe population responses. First, single neurons have receptivefields with a substantial spatial extent that increases rapidlyas a function of retinal eccentricity (Hubel and Wiesel, 1974;Van Essen et al., 1984). Second, the response amplitude ofsingle neurons increases nonlinearly with contrast, typicallyreaching response saturation at low to modest contrasts(Albrecht and Hamilton, 1982). Third, the tuning of single neuronsis typically invariant with contrast, even in the saturated responserange (Albrecht and Hamilton, 1982; Albrecht and Geisler, 1991;Heeger, 1991, 1992). Fourth, the latency of the response ofsingle neurons decreases as a function of stimulus contrast(Dean and Tolhurst, 1986; Carandini and Heeger, 1994; Albrecht,1995). Although these properties are common in V1 neurons,there is a vast heterogeneity among the neurons, and thus it isunclear how these properties are combined and manifested atthe population level. In addition, single-unit and multiple-unitstudies in V1 have focused mainly on responses at or nearthe center of activity produced by the stimulus. Responses atlocations more peripheral to the center of activity are largelyunknown.Here we provide a complete quantitative description of thereal-time spatiotemporal dynamics of V1 population responsesto a small, briefly presented (200 ms), localized stationary visualstimulus. Most measurements of response properties in V1 havebeen performed using drifting stimuli with relatively long dura-tions (several seconds) to approximate a steady-state condition.However, natural saccadic inspection of a visual scene typicallyproduces transient stimulation: 200–300 ms fixations separatedby rapid eye movements. In addition, although it is commonto analyze cortical responses by their peak responses andlatencies (phases) for drifting stimuli, the falling edges of theresponses can potentially provide useful information for brieflypresented stimuli (Bair et al., 2002). Thus, to fully understandthe properties of the population responses under natural condi-tions, it is important to measure the complete time courses ofresponses to briefly presented stimuli.We used voltage-sensitive dye imaging (VSDI; Grinvald andHildesheim, 2004) in alert, fixating monkeys, to measure popula-tion responses in the superficial layers of macaque V1 over anarea of approximately 1 cm2. The imaged area covered the entireregion activated by the small local stimulus. We found severalunexpected properties that are not obvious from single unitresponses. First, the spatial profile of the peak response is inde-pendent of stimulus contrast. Second, responses start to rise atall locations approximately at the same time, but rise at a fasterrate at the center of activity than at peripheral locations. Third,both the latency and steepness of the rising edge of the responsedepend on stimulus contrast. Finally, after stimulus offset, theNeuron 64, 943–956, December 24, 2009 ª2009 Elsevier Inc. 943responses at all locations fall simultaneously and at the samerate,