Abstract The responses of 30 V1 complex cells wererecorded using a complete set of transiently presented,oriented stimuli of different contrasts. A back-propaga-tion neural network was used to predict the multivariatevisual stimuli from the neuronal responses on a trial-by-trial basis. For single neurons, the strength of the re-sponse was much better at predicting the orientation of avisual stimulus than its contrast. Using the temporalmodulation of the response improved the ability to pre-dict the contrast of a stimulus without affecting the abili-ty to predict the orientation. Removing stimulus latencyfrom the responses, by time-shifting the individual re-sponses an amount equal to the average latency, signifi-cantly reduced the ability to predict stimulus contrast,demonstrating that the response latency is reliableenough, even for a single neuron and a single trial, for itto be used to help determine stimulus contrast. Poolingthe responses from a group of 11 neurons demonstratedthat small groups of neurons can accurately representmultivariate stimuli in a single trial.Key words Latency · Temporal Coding · Population · Ensemble · TransientIntroductionIt has been previously determined that V1 complex cellsencode significant amounts of stimulus-related informationin the temporal modulation of their responses (Richmondand Optican 1990). It has also been found that the inherentambiguity of encoding for multivariate visual stimuli withthe responses of a single neuron can be reduced by usingthe temporal modulation (Gawne et al. 1991). More recent-ly, it has been shown that, while the response strength isprimarily determined by stimulus orientation, response la-tency is more strongly a function of the stimulus contrast(Gawne et al. 1996b). [Studies using drifting gratings haveshown response-phase shifts with changes in stimulus con-trast without concomitant changes in response strength,which seems likely to be another aspect of this phenome-non (Albrecht 1995)]. It has also been shown that there issignificant temporal coding for stimulus contrast in the re-sponse of a complex cell for transiently presented stimuli(Mechler et al. 1998), although this latter study did not co-vary different stimulus parameters or attempt to separateout the effects of the different aspects of the temporal vari-ation of the response.The purpose of this study was to determine how ef-fectively the responses of both individual and smallgroups of V1 complex cells can represent stimulus orien-tation and contrast when these parameters vary simulta-neously. Stimuli consisting of bars of either eight or 12uniformly distributed orientations were presented ateight different contrasts, flashed on the receptive fieldsof V1 complex cells of awake rhesus monkeys (Macacamulatta). The responses were quantified as either the av-erage firing rate or as the coefficients of the principalcomponents of the smoothed post-stimulus histograms (a measure of the temporal variation). The responseswere used to predict the stimuli using a neural networktrained by a standard back-propagation algorithm. Byvarying the aspects of the neuronal responses used topredict the stimuli, and by pooling the responses of mul-tiple neurons, limits were set on the ability of single andsmall numbers of neurons to encode for more than onestimulus parameter at the same time.The major conclusions to be drawn from this studyare: (1) that the ability of single V1 neurons to encodeinformation about contrast is primarily a function of thetemporal modulation of the response; (2) that responselatency is a large although, not exclusive component ofthe useful temporal modulation; (3) that, by using thetemporal modulation of the response, even a single neu-ron can unambiguously encode for both orientation andcontrast in its response; (4) that small numbers of V1T.J. Gawne (✉)Department of Physiological Optics, University of Alabama at Birmingham, 924 South 18th St., Birmingham, AL 35294, USAe-mail: [email protected].: +1-205-9345725, Fax: +1-205-9345725Exp Brain Res (2000) 133:293–302Digital Object Identifier (DOI) 10.1007/s002210000381RESEARCH ARTICLETimothy J. GawneThe simultaneous coding of orientation and contrast in the responses of V1 complex cellsReceived: 9 September 1999 / Accepted: 12 February 2000 / Published online: 11 April 2000© Springer-Verlag 2000complex cells with complementary orientation tuningcan encode for multivariate stimuli with a high degree ofaccuracy; and (5) that, at least under some conditions,pooling the responses can partially overcome some ofthe limitations of a spike-count code, but only for limitedconditions.Materials and methodsRecordings from single V1 complex neurons were made in twoawake rhesus monkeys (Macaca mulatta). The monkeys were firstanesthetized with halothane, and a recording chamber (Crist Instru-ments) was implanted over area V1. Additionally, a coil of Teflon-coated stainless-steel wire (Cooner Wire) was implanted under Ten-on’s capsule of one eye so that the magnetic field/search coil meth-od (Robinson 1963; Judge et al. 1980) (Riverbend Instruments)could be used to monitor eye position. Upon recovery, the monkeyswere trained to fixate a spot of light to within ±0.3° of a 0.04°square fixation point and given juice rewards for maintaining fixa-tion. The monkeys maintained a better fixation than the eye-positionlimits, with an average deviation from the fixation point of ±8 minof arc (SD). The ability of primates to accurately fixate in simpletasks has been demonstrated before (Snodderly and Kurtz 1985).There is evidence that the errors of fixation in awake monkeyscan affect the responses of visual-cortical neurons (Gur andSnodderly 1987; Gur et al. 1997). However, previous studies un-der similar conditions have determined that this variance of eyeposition has little effect upon the responses of V1 complex cells(at least for receptive fields of this eccentricity) (Kjaer et al.1997), which is not surprising given the long-known insensitivityof these cells to small shifts in retinal position (Hubel and Wiesel1968). Given the high precision of fixation for a simple staticviewing task, the parafoveal location, and the demonstrated insen-sitivity of these neurons to small shifts in stimulus position, ittherefore seems unlikely that the small changes in eye position ofthese continuously fixating monkeys should significantly bias theresults of these experiments.Glass-coated platinum-iridium