Visual Neuroscience ~2007!, 24, Page 1 of 12. Printed in the USA. Copyright © 2007 Cambridge University Press 0952-5238007 $25.00 DOI: 10.10170S0952523807070563

Temporal properties of surround suppression in cat primary visual cortex

SÉVERINE DURAND, TOBE C.B. FREEMAN,1 and MATTEO CARANDINI 2 Institute of Neuroinformatics, University of Zurich and Swiss Federal Institute of Technology, Zurich, Switzerland (Received January 12, 2007; Accepted May 29, 2007!

Abstract The responses of neurons in primary visual cortex ~V1! are suppressed by stimuli presented in the region surrounding the receptive field. There is debate as to whether this surround suppression is due to intracortical inhibition, is inherited from lateral geniculate nucleus ~LGN!, or is due to a combination of these factors. The mechanisms involved in surround suppression may differ from those involved in suppression within the receptive field, which is called cross-orientation suppression. To compare surround suppression to cross-orientation suppression, and to help elucidate its underlying mechanisms, we studied its temporal properties in anesthetized and paralyzed cats. We first measured the temporal resolution of suppression. While cat LGN neurons respond vigorously to drift rates up to 30 Hz, most cat V1 neurons stop responding above 10–15 Hz. If suppression originated in cortical responses, therefore, it should disappear above such drift rates. In a majority of cells, surround suppression decreased substantially when surround drift rate was above ;15 Hz, but some neurons demonstrated suppression with surround drift rates as high as 21 Hz. We then measured the susceptibility of suppression to contrast adaptation. Contrast adaptation reduces responses of cortical neurons much more than those of LGN neurons. If suppression originated in cortical responses, therefore, it should be reduced by adaptation. Consistent with this hypothesis, we found that prolonged exposure to the surround stimulus decreased the strength of surround suppression. The results of both experiments differ markedly from those previously obtained in a study of cross-orientation suppression, whose temporal properties were found to resemble those of LGN neurons. Our results provide further evidence that these two forms of suppression are due to different mechanisms. Surround suppression can be explained by a mixture of thalamic and cortical influences. It could also arise entirely from intracortical inhibition, but only if inhibitory neurons respond to somewhat higher drift rates than most cortical cells. Keywords: Primary visual cortex, Lateral geniculate nucleus, Suppression, Temporal frequency, Adaptation

the test centered on the receptive field ~Blakemore & Tobin, 1972; Maffei & Fiorentini, 1976!. This effect is known as “surround suppression” and has also been called “end-stopping,” “size tuning,” and “hypercomplexity” ~Fitzpatrick, 2000!. Surround suppression is highly selective for orientation, being strongest when the mask is at the cell’s preferred orientation ~DeAngelis et al., 1994; Sengpiel et al., 1997!. In the earliest explanation, these two suppression phenomena result from inhibition between cortical neurons ~Hubel & Wiesel, 1965; Morrone et al., 1982!. According to this explanation, crossorientation suppression would originate from a pool of cortical neurons selective for a broad range of orientations, whose receptive fields would overlap with that of the recorded neuron ~Fig. 1A!. Similarly, surround suppression would originate from cortical neurons selective for the same orientation as the recorded neuron with receptive fields covering a broader region of visual space ~Fig. 1B!. Inhibition could operate within V1 ~Hubel & Wiesel, 1965; Blakemore & Tobin, 1972!, or involve feedback from higher cortical areas ~Allison et al., 2001; Angelucci et al., 2002; Cavanaugh et al., 2002a; Levitt & Lund, 2002; Angelucci & Bullier, 2003; Angelucci & Bressloff, 2006!.

Introduction The response of a neuron in primary visual cortex to test stimuli in its receptive field can be suppressed by adding mask stimuli that evoke little response when presented alone. Two broad classes of visual suppression have been distinguished. The first is observed by superimposing a mask onto the test, within the receptive field ~Morrone et al., 1982; Bonds, 1989; DeAngelis et al., 1992!. This form of suppression can be obtained with masks of all orientations, including the orientation orthogonal to the preferred orientation of the cell. It is known as “cross-orientation suppression,” and has also been called “cross-orientation inhibition.” The second form of suppression is found when adding a mask in the area surrounding

Address correspondence and reprint requests to: Séverine Durand: Riken, Brain Science Institute, Laboratory for Neuronal Circuit Development, 2-1 Hirosawa, Wakoshi, Saitama, 351-0198, Japan. E-mail: severine@ brain.riken.jp. 1 Current address: Genedata AG, Maulbeerstrasse 46, CH-4016 Basel, Switzerland. 2 Current address: Smith-Kettlewell Eye Research Institute, 2318 Fillmore Street, San Francisco, CA 94115.

1

2

Fig. 1. Stimuli used to probe suppression, and interpretation of results based on the intracortical hypothesis. ~A! Cross orientation suppression. Stimuli are composed of a test grating and an orthogonal mask grating. The intracortical hypothesis ~top! ascribes cross-orientation suppression to inhibition between cortical cells with overlapping receptive fields and different orientation preferences. ~B! Surround suppression. Stimuli are composed of a test grating and a surrounding mask grating with the same orientation. The intracortical hypothesis ~top! ascribes surround suppression to inhibition between neighboring cortical cells with similar receptive fields.

S. Durand et al. some surround suppression is seen already seen in the responses of LGN neurons ~Solomon et al., 2002; Ozeki et al., 2004; Webb et al., 2005b!, where it might be caused by mechanisms of contrast gain control that originate as early as in retina ~Bonin et al., 2005!. It seems increasingly likely that surround suppression is due to a combination of factors ~see Discussion!: inheritance of suppression from the LGN inputs, local inhibitory circuits within V1, and feedback from higher visual areas. To further elucidate the sources of surround suppression and the differences between surround suppression and cross-orientation suppression, we made a side-by-side comparison between the two forms of suppression. We made these measurements in cat, which is the species where suppression has been most thoroughly investigated. First, we measured the ability of masks of different temporal frequencies to cause suppression. We thus estimated the temporal resolution of surround suppression and compared it with that of cross-orientation suppression on one hand, and with that of LGN and V1 neurons on the other. Then, we asked if we could reduce surround suppression through prolonged adaptation to annulus stimuli. These experiments yielded radically different outcomes for the two forms of suppression, and emphasize their different mechanisms and sources. Partial analyses of these data have appeared in conference presentations ~Durand et al., 2001, 2002! and in a related paper ~Freeman et al., 2002!.

Materials and methods However, there is now substantial evidence that inhibition is not the correct explanation for cross-orientation suppression. This evidence derives from a comparison between the visual properties of cross-orientation suppression and that of cortical neurons. Crossorientation suppression is observed at substantially higher drift rates than those eliciting responses in most neurons in area 17 or 18 ~Allison et al., 2001; Freeman et al., 2002; Li et al., 2005; Sengpiel & Vorobyov, 2005!. In fact, the temporal resolution of cross-orientation suppression resembles more that of afferents from lateral geniculate nucleus ~LGN!, which typically respond to drift rates of 20 Hz and higher ~Orban et al., 1985; Saul & Humphrey, 1990!. In addition, cross-orientation suppression is largely immune to adaptation, the marked and selective reduction in cortical responsiveness that follows prolonged visual stimulation ~Maffei et al., 1973; Albrecht et al., 1984; Ohzawa et al., 1985; Carandini & Ferster, 1997; Sanchez-Vives et al., 2000!. If the suppressive signals originated in cortex, presenting a mask stimulus for a long period of time should reduce the cells’ subsequent response to it and consequently the suppression that it provides. Cross-orientation suppression, instead, is immune to adaptation ~Freeman et al., 2002; Li et al., 2005; Sengpiel & Vorobyov, 2005!. These properties suggest that cross-orientation suppression originates from LGN signals ~Freeman et al., 2002!. It may be due to saturation in LGN responses ~Li et al., 2006; Priebe & Ferster, 2006!, with a possible contribution of depression at the thalamocortical synapse ~Freeman et al., 2002!. There is also some evidence that the inhibition hypothesis may not be the correct explanation to surround suppression. For example, in primates surround suppression can be seen with stimuli of very low spatial frequency, which are ineffective in driving visual cortex ~Webb et al., 2005a!. Moreover, pharmacological manipulations that reduce the strength of GABA inhibition do not affect the impact of surround suppression ~Ozeki et al., 2004!. In fact,

Methods for the physiological preparation are standard in our laboratory ~Freeman et al., 2002!. Briefly, adult cats were initially anesthetized with ketamine Hcl ~20 mg0kg! and acepromazine ~0.1 mg0kg!, and premedicated with atropine sulfate ~0.05 mg0kg!. Anesthesia was subsequently maintained throughout the course of the experiment by a continuous IV infusion of sodium Penthotal ~1– 4 mg0kg0h!. Paralysis was induced with pancuronium bromide ~0.15 mg0kg0h! and artificial respiration maintained with a mixture of N2O and O2 ~commonly 2:1!. The corneas were protected by contact lenses and appropriate corrective lenses were set in front of each eye. EEG, ECG, temperature and end-tidal CO2 were monitored continuously. A craniotomy was performed centered around Horsley-Clarke coordinates 2.5 mm lateral and 7 mm posterior. V1 neurons were in area 17 or at the 17018 border, with receptive fields usually within ;58 of eccentricity. Extracellular signals were recorded using glass-coated tungsten microelectrodes or quartz-coated Platinum0Tungsten microelectrodes, sampled at 12 kHz, and stored for offline spike discrimination. The veterinary office of Canton Zurich approved all procedures. Visual stimuli were drifting sinusoidal gratings enclosed in circular patches and annuli, displayed monocularly using the Psychophysics Toolbox ~Matlab environment! ~Brainard, 1997; Pelli, 1997! on a calibrated monitor ~Sony Multiscan G-500, mean luminance 39 cd0m 2, refresh rate 120–160 Hz!. In a set of standard measurements, we established optimal orientation, spatial frequency, drift rate, position and size of drifting gratings at 50% contrast. The optimum size of the patch for subsequent experiments was decided after presenting stimuli of increasing sizes in random orders and finding the size that gave the maximal response. We classified V1 cells as simple or complex based on the “relative modulation index” ~F10F0! measured at the optimal spatial frequency ~Skottun et al., 1991!. Stimuli lasted for 1–5 s and were

Surround suppression

3

randomly interleaved within each block to minimize the effects of adaptation. Usually we repeated each block three times. We then measured the strength of suppression. These measurements involved two gratings, a test, and a mask. The test drifted in the cell’s preferred direction and drift rate, and was windowed by a disk of optimal size. For measurements of cross-orientation suppression, the mask was superimposed on the test, and drifted in an orthogonal direction. The mask had one of 6–8 drift rates between 0.2 and 24 Hz. For measurements of surround suppression, the mask was enclosed in an annulus surrounding the test, and drifted in the same direction as the test. The dimensions of the annulus were determined by a preliminary experiment where single annuli were presented with different inner diameters. We chose the inner diameter of the annulus so that the annulus alone would elicit minimal responses ~generally less than 10% of the maximal response!. This inner diameter was always larger than the diameter of the test patch and we held the gap between annulus and patch at the mean luminance of the screen. The mask had one of 6–8 drift rates between 1 and 21 Hz. In both experiments, the test contrast took one of 5 or 6 values between 0 and 100%, and the mask contrast was either 0% or 50%. Blocks where we measured cross-orientation suppression and blocks where we measured surround suppression were separate. The control condition ~test alone!, was measured twice, and intermixed with both blocks. For adaptation experiments, the contrast of each grating was varied independently, producing a total of 26 stimuli. Stimuli lasted 2– 4 s and were presented in randomized order. Adaptation stimuli were presented initially for 30 s and then for 3– 6 s prior to the presentation of each probe stimulus. Neurons were first adapted to a blank screen ~a control experiment!, then to the annulus stimulus displayed at 50% contrast. In a few neurons, this experimental order was reversed. Adaptation experiments were not performed on the same cells as measurements of temporal preferences of suppression. We describe the effects of suppression by fitting a model to the responses ~Freeman et al., 2002!. The first stage of the model is linear in contrast, whereas the second stage incorporates a divisive suppressive signal. The response of the first stage to a grating is a constant B plus a sinusoid with amplitude A and phase P at the frequency v of the stimulus, multiplied by stimulus contrast c. For test and mask gratings, thus, these responses are L test ~t ! ⫽ ctest @Btest ⫹ A test sin~2pvtest t ⫹ Ptest !# L mask ~t ! ⫽ cmask @Bmask ⫹ A mask sin~2pvmask t ⫹ Pmask !# . The output of the full model to the superposition of test and mask is given by: R~t ! ⫽

@L test ~t ! ⫹ L mask ~t !# n 1 ⫹ ~dtest ctest ! n ⫹ ~dmask cmask ! n

,

~1!

where the exponent n approximates the nonlinear effects of the encoding of membrane potential into firing rate, and dtest and dmask indicate the effectiveness of test and mask in suppressing the responses. The model can be applied to both simple cells ~where A test ⬎ Btest ! and to complex cells ~where Btest ⱖ A test !. For a typical experiment, the mask does not elicit much of a response by itself ~L mask ⬇ 0!. In this case, the model responses averaged over time can be approximated by a simple expression of test contrast ctest , a hyperbolic ratio:

R~c O test ! ⬇ R max

n ctest n c50

n ⫹ ctest

,

where R max is maximal response and c50 is semi-saturation contrast, at which the responses to test alone reach half of the asymptotic value ~Albrecht & Hamilton, 1982!. The model predicts suppression because the semi-saturation contrast grows with mask contrast: n c50 ⫽

n n cmask ! ~1 ⫹ dmask n dtest

.

The only factor in the model that causes suppression is the parameter dmask . If dmask were zero, there would be no suppression. We could not estimate dmask directly because it trades with other parameters ~in particular, one can divide Eq. ~1! by the factor dtest and obtain the same results!. We could, however, estimate measures that are proportional to dmask and are not corrupted by the choice of other parameters. In experiments where we measured the temporal resolution of suppression, we only tested two values of cmask , 0, and 50%, but we could safely assume that the parameter dtest was constant. We therefore used the following suppression index, the ratio of c50 in the presence of the annulus and in the absence of the annulus: S ⫽ c50 @disk ⫹ annulus#0c50 @disk#. This index of suppression is an intuitive value; it is the factor by which disk contrast should be multiplied in the presence of the annulus to obtain the same response as when the disk is alone. Its expression is proportional to dmask , because, at 50% mask contrast, n it equals ~1 ⫹ dmask 0.5 n !10n . For experiments where we measured the susceptibility of suppression to adaptation, instead, we did test multiple values of cmask but we could not assume that R max , dtest or any other parameter stayed constant across adaptation conditions. What matters here is the degree to which dmask is changed relative to dtest . For these experiments, therefore, we used a more robust but less intuitive measure of suppression ~Freeman et al., 2002!, the suppression index k ⫽ dmask 0dtest . Crucially, though we used different measures of suppression in our two sets of experiments, we used the very same measures when comparing one form of suppression to the other. We performed the fits by minimizing the mean square error between model prediction and cell firing rate. Firing rate was obtained by low-pass filtering the spike trains with a Gaussian running window ~s ⫽ 25 ms! and by averaging across block repeats. We assessed fit quality by considering the percentage in the variance of the mean responses that was explained by the model. If m j and rj are the predicted response and the mean of the responses to the j-th stimulus, and r0 is the overall mean of the responses, the percentage of the variance explained by the model is V ⫽ 100@1 ⫺ S j ~m j ⫺ rj ! 20S j ~rj ⫺ r0 ! 2 # .

4

S. Durand et al.

Fig. 2. Surround suppression in two V1 neurons. ~A! Visual stimuli were drifting gratings enclosed in a disk window, in the absence ~top! and in the presence ~bottom! of a grating in a surrounding annulus. Disk contrast was varied randomly, from 0 ~left! to 50% ~right!. ~B! Responses of a complex cell ~unit 32.4.3, expt. 15.11!. Histograms indicate average response to the passage of a bar of the disk grating. Curves represent fits by the gain control model described in Methods. ~C! Same, for a simple cell ~unit 16.2.9, expt. 21.25!.

To describe the dependence of various quantities on drift rate, v, we fit the measurements with a simple descriptive function: the sum of two truncated Gaussians ~Freeman et al., 2002!. The truncated Gaussians meet at the preferred frequency vtop . The Gaussian has standard deviation s ⫽ slow for frequencies v ⬍ vtop , and s ⫽ shigh for frequencies v ⬎ vtop . Because a Gaussian reaches half of its maximal amplitude at ln~2!102 s, the ~50% high! cutoff frequency is vtop ⫹ ln~2!102 shigh .

can be observed by plotting the mean response as a function of contrast. This is done in Fig. 3A for the complex cell and in Fig. 3B for the simple cell. When the disk is presented alone, the mean responses increase sigmoidally with disk contrast. Increases

Results Surround suppression is readily demonstrated by putting an annulus stimulus around an optimal grating presented in the receptive field ~Fig. 2!. V1 neurons give robust responses to gratings enclosed in an optimally sized disk centered on their receptive field ~Figs. 2B, top and 2C, top!. In the presence of a grating in a surrounding annulus, however, these responses are suppressed ~Fig. 2B, bottom and Fig. 2C, bottom!. To analyze these responses, we fit them with a descriptive model, a function of disk and annulus contrast ~see Materials and methods!. This model is formally identical to one we have employed to describe suppression within the receptive field ~Freeman et al., 2002!. The model is divisive, with a numerator growing linearly with contrast, and a denominator that exerts suppression. The numerator varies in time, and is the sum of a sinusoid that modulates at the frequency of the stimulus ~to account for simple cell responses to gratings! and a constant ~to account for complex cell responses!. The denominator is constant in time, and is essentially a weighted sum of disk and annulus contrast. Divisive models of this kind have been shown to provide good fits to responses to gratings in the receptive field ~Carandini et al., 1997; Freeman et al., 2002! and in the surrounding regions ~Chen et al., 2001; Cavanaugh et al., 2002a!. Indeed, the model does a good job at describing responses both in the presence and in the absence of the annulus ~Figs. 2B, 2C, thick curves!. On average it explained 87.9 6 9.0 % ~s.d., N ⫽ 56! of the variance in the mean responses. This percentage was ⬍70% in only 3056 cells, ⬍80% in 11056 cells. In the following, we concentrated on the majority of cells ~53056! in which the model explained ⬎70% of the variance. In these cells the model can be safely used to describe the responses and the effect of surround suppression. Surround suppression constitutes an arithmetical division ~Carandini et al., 1997; Sengpiel et al., 1998; Chen et al., 2001; Sceniak et al., 2001; Cavanaugh et al., 2002a; Carandini, 2004!. This effect

Fig. 3. Effect of surround suppression on contrast responses. Mean responses of the two cells seen in Fig. 2, as a function of disk contrast. Rows correspond to a different annulus drift rates. Clear symbols are responses to disk alone ~reported only on top row to reduce clutter!. Dark symbols are responses in the presence of the annulus. Curves are fits of the model. ~A! Responses of the example complex cell. Data of Fig. 2B appears in A4 Disk drift rate was 5 Hz. ~B! Responses of the example simple cell. Data of Fig. 2C appears in B3 . Disk drift rate was 6 Hz. Both examples show a rightward shift of the response curve when the annulus is added to the disk, consistent with a divisive effect of surround suppression. This suppression is reduced at high annulus drift rates.

Surround suppression in annulus contrast shift the sigmoid to the right. Because the scale in the abscissa is logarithmic, it is as if the annulus had divided the test contrast seen by the cell. A similar effect is seen with crossorientation suppression ~e.g., Carandini et al., 1997; Sengpiel et al., 1998; Freeman et al., 2002!. The model provides a good summary of the effects of surround suppression ~Fig. 3, curves!. For a typical experiment, in which the annulus hardly elicits any response by itself, model responses are approximated by a simple function of disk contrast cdisk , the hyperbolic ratio. This is the function that is commonly fitted to contrast responses ~Albrecht & Hamilton, 1982!. Its key parameter is semi-saturation contrast c50 , the contrast at which responses reach half of their asymptotic value. In our model, semi-saturation contrast depends on stimulus configuration ~Materials and methods!, being larger in the presence than in the absence of the annulus. The model captures the effect of surround suppression through an increase in the semi-saturation contrast c50 ~Heeger, 1992; Freeman et al., 2002!. For example, for the complex cell in Fig. 3A4 semi-saturation contrast c50 goes from 43% in the absence of the annulus, to 69% in the presence of the annulus. Similarly, for the simple cell in Fig. 3B4 , semi-saturation contrast c50 is 20% for the disk alone, and 69% in the presence of the annulus. Incidentally, all these values of semi-saturation contrast are larger than those reported in earlier studies ~Albrecht & Hamilton, 1982!, most likely because of the small size of our disk stimuli. Because V1 neurons are more selective for stimulus size at high contrast than at low contrast ~Jagadeesh & Ferster, 1990; Kapadia et al., 1999; Sceniak et al., 1999, 2001; Cavanaugh et al., 2002a!, their responses saturate at higher contrasts for small stimuli than for large stimuli. Suppression with fast gratings To investigate the temporal resolution of surround suppression, we recorded from 68 V1 neurons. Results reported here concern 56 of these cells, which fulfilled the following requirements: ~1! the maximum average response to disk alone was greater than 5 spikes0s ~7068 cells excluded!; ~2! the ratio of the responses to annulus alone and disk alone was ⬍0.25 ~5061 cells excluded!. We imposed the first requirement because we study reductions in response, so we need somewhat sizeable responses to begin with. We imposed the second requirement because our data are more easily understood if the annulus is far enough from the receptive field center so as not to cause much response by itself. As mentioned before, we also remove three other cells where the model was not describing the data in a satisfactory way. The strength of suppression depends on the drift rate of the grating in the annulus. As explained in Materials and methods, for these experiments we measured the strength of suppression by estimating the degree to which the mask increases the semisaturation contrast c50 . For our example complex cell ~Figs. 3A1 to A6 !, semi-saturation contrast is low ~c50 ⫽ 43%! when annulus drift rate is low ~1 Hz, Fig. 3A1 !, maximal ~c50 ⫽ 69%! when annulus drift rate is 8 Hz ~Fig. 3A4 !, and low again ~c50 ⫽ 39%! at high annulus drift rate ~21 Hz! ~Fig. 3A6 !. Similarly, for the simple cell, c50 is low ~39%! when the annulus drifts slowly ~Fig. 3B1 !, maximal ~c50 ⫽ 69%! for an annulus drift rate of 8 Hz ~Fig. 3B4 !, and becomes low again ~c50 ⫽ 36%! when the annulus drift rate is 21 Hz ~Fig. 3B6 !. The dependence of the strength of suppression on annulus drift rate is illustrated in Fig 4. For the example complex cell ~Fig. 4A!, suppression is maximal ~S ⫽ 1.9! when the drift rate of the annulus

5

Fig. 4. Temporal tuning of suppression. Data shows the strength of suppression for the cells seen in Figs. 2 and 3, as a function of annulus drift rate. Symbols are indexes of suppression S. Curves are fits with a simple descriptive function ~see Materials and methods!. ~A! Tuning of suppression for the example complex cell. The arrow indicates the high drift rate cutoff: 14.4 Hz. ~B! Tuning of suppression for the example simple cell. High drift rate cutoff is 19.1 Hz. In both cases ~A and B!, the curves reach their maximum for an annulus drift rate of 8 Hz and decrease at higher values.

~2 Hz or 8 Hz! is near the preferred rate of the cell ~5 Hz! and diminishes dramatically ~S ⫽ 1.1! for annulus drift rate of 21 Hz. The same observation is made for our example simple cell ~Fig. 4B!. The maximum suppression ~S ⫽ 3.4! is seen when the annulus drifts at 8 Hz, whereas when it drifts at 21 Hz, suppression drops substantially ~S ⫽ 1.8!. To summarize these temporal frequency profiles, we fitted them with a simple descriptive function ~Fig. 4, curves!. We concentrated on 46053 cells that showed clear surround suppression ~maximal index of suppression S ⬎ 1.3!, and found that the descriptive function is appropriate for the task. The function explained 83.7 6 2.4% of the variance ~s.d., N ⫽ 46!. For comparison, the function explained 84% of the variance for the example complex cell ~Fig. 4A!, and 99% of the variance for the example simple cell ~Fig. 4B!. Because we intend to describe data using the function, we excluded from further analysis a few cells ~5046! where the descriptive function accounted for ⬍50% of the variance in the data. We defined the high drift rate cutoff of these tuning curves as the drift rate of the annulus for which suppression is 50% of its maximum. For our two example cells in Figs 4A and 4B, cutoff is 14.4 Hz and 19.1 Hz ~Fig. 4, arrows!. These cutoffs are lower than the fastest annulus that we tested ~21 Hz!. Most of the cells had high drift rate cutoffs situated around these values ~Fig. 5F!. The majority of cells showed cutoffs between 6 and 20 Hz ~29041 cells!. Only 9041 cell displayed cutoffs that were higher than 20 Hz. All cutoffs above 21 Hz were conservatively set to this value since it is the fastest surround drift rate used in the experiment. The mean of cutoffs for the complete population of cells is 15.5 6 7.0 Hz, with a median at 15.5 Hz.

Temporal resolution of LGN and V1 neurons The difference in temporal resolution between cross-orientation suppression and surround suppression reminds of the difference in temporal resolution between geniculate and cortical neurons ~Orban et al., 1985!. In cats, V1 neurons barely respond to drift rates above 10–15 Hz ~Ikeda & Wright, 1975; Movshon et al., 1978;

6

S. Durand et al.

Fig. 5. Distributions of preferred and cutoff drift rate for two forms of suppression and for the responses of V1 and LGN neurons. Suppression was measured from the index of suppression S. Responses were measured from mean firing rate in response to drifting gratings. ~A! Preferred drift rate for cross-orientation suppression ~n ⫽ 35!. ~B! Preferred drift rate for surround suppression measurements ~n ⫽ 41!. ~C! Preferred drift rate for the responses of V1 neurons ~n ⫽ 294!. ~D! Preferred drift rate for the responses of LGN neurons ~n ⫽ 86!. For A, B values above 20 Hz are grouped in the last bin of each histogram. For C, D values above 30 Hz are grouped in the last bin of each histogram. White bars indicate values that were extrapolated. ~E–H! Similar measurements for cutoff drift rate.

Saul & Humphrey, 1992!, whereas LGN neurons give strong responses to these frequencies ~Lehmkuhle et al., 1980; Orban et al., 1985; Saul & Humphrey, 1990!. To facilitate a comparison between suppression and responses, we measured the preferred drift rate of V1 and LGN cells ~Figs. 5C and 5D!. Our sample included 86 LGN and 294 V1 neurons. LGN and V1 cells were mostly recorded from different animals. Stimuli were high contrast gratings varying in drift rate. We then fitted responses with the same descriptive function as in Fig. 4, and used them to derive preferred and cutoff drift rates. As expected, the difference in preferred drift rate between V1 and LGN cells is substantial ~Saul & Humphrey, 1990, 1992!. The average preferred drift rate for V1 cells ~Fig. 5C! is 4.1 6 4.0 Hz, much lower than that for LGN cells ~Fig. 5D!, which is 14.9 6 9.5 Hz ~14.8 6 9.7 Hz excluding extrapolated values shown in white!. A similar difference between LGN and V1 is observed in the cutoff drift rate, which is much lower in V1 than in LGN ~Figs. 5G

and 5H!. The analysis of the above population reveals that the cutoff drift rate of V1 neurons lies between 1 and 20 Hz ~Fig. 5G!, being on average 9.0 6 8.4 Hz ~7.7 6 3.3 Hz, without extrapolation!. The median cutoff was 7.5 Hz ~7.3 Hz, without extrapolation!. Only 3 cells out of 294 had cutoffs higher than 30 Hz, and 9 out of 294 had cutoffs above 20 Hz. LGN cells have higher cutoff drift rates ~Freeman et al., 2002; Li et al., 2005!: The majority of LGN neurons have a cutoff drift rate above 15–20 Hz ~Fig. 5H!. Cutoffs were higher than 30 Hz in 41086 cells and higher than 20 Hz in 58086 cells. The average cutoff for LGN cells responses was 27.6 6 12.1 Hz ~23.5 6 10.5 Hz, without extrapolation!, with a median at 29.4 Hz ~23.5 Hz, without extrapolation!. Comparison of temporal preferences We can now compare the temporal resolution of cross-orientation and surround suppression with that of LGN and V1 neurons. This

Surround suppression comparison can shed light on the origin of the suppressive signals. Indeed, recent studies of cross-orientation suppression have revealed that the suppressive signals are responsive to high temporal frequencies ~Allison et al., 2001; Freeman et al., 2002; Li et al., 2005!, leading to the suggestion that they originate in LGN responses. It is of interest to follow the same approach for surround suppression and to contrast the results obtained in the two forms of suppression. As expected from previous studies ~Freeman et al., 2002; Li et al., 2005!, the temporal preferences of cross-orientation suppression resemble those of LGN responses more than those of V1 responses. Indeed, the distribution of preferred drift rates ~Fig. 5A! resembles that observed for LGN responses ~Fig. 5D! more than that observed for V1 responses ~Fig. 5C!. Indeed, the average preferred drift rate was 11.1 Hz for cross-orientation suppression, 14.9 Hz for LGN responses, and only 4.1 Hz for V1 responses. A similar argument can be made for the cutoff drift rate, the highest frequency that gives rise to 50% of maximal response or maximal suppression. As observed in earlier studies ~Allison et al., 2001; Freeman et al., 2002; Li et al., 2005!, the distribution of cutoff drift rates for cross-orientation suppression ~Fig. 5E! resembles that observed for LGN responses ~Fig. 5H! more than that observed for V1 responses ~Fig. 5G!. For instance, the average cutoff drift rate was 21.7 Hz for cross-orientation suppression, 27.6 Hz for LGN responses, and only 9.0 Hz for V1 responses. In summary, at high drift rates cross-orientation suppression and LGN responses are strong, but responses in V1 are weak. The temporal preferences of surround suppression, instead, are not radically different from those of V1 neurons; they seem to be intermediate between those of LGN neurons and those of V1 neurons. Indeed, the distribution of preferred drift rates for surround suppression ~Fig. 5B! contains higher values than those seen in V1 responses ~Fig. 5C! but somewhat lower values than those seen in LGN responses ~Fig. 5D!. The average preferred drift rate for surround suppression was 7.2 Hz, midway between 4.1 Hz for V1 responses, and 14.9 Hz for LGN responses. Likewise, the distribution of cutoff drift rates for surround suppression ~Fig. 5F! is intermediate between those observed for V1 responses ~Fig. 5G! and for LGN responses ~Fig. 5H!. The average cutoff drift rate for surround suppression was 15.5 Hz, midway between 9.0 Hz for V1 responses, and 27.6 Hz for LGN responses. The temporal properties of suppression by the annulus stimuli did not depend on whether the annulus alone could elicit a response. Indeed, if we restrict the analysis to the 36 cells in which the annulus elicited ⬍10% of the maximal response ~instead of ⬍25%!, we find that surround suppression had a mean preferred drift rates of 6.8Hz ~instead of 7.2 Hz! and a mean cutoff drift rates of 14.7 Hz ~instead of 15.5! Hz. We confirmed that the temporal properties of cross-orientation suppression resemble those of LGN responses by comparing the median values of the distributions and assessing the significance of their deviations with bootstrap tests ~Efron & Tibshirani, 1993!. We could not reject the hypothesis that the signals causing crossorientation suppression are the same that cause LGN responses: the probability that the two distributions are the same is p ⫽ 0.08 based on the high-frequency cutoffs, and p ⫽ 0.52 based on the preferred temporal frequencies. Conversely, we could reject identity ~at the p ⫽ 0.01 level! for all other compared pairs of distributions seen in Fig. 5. These results resemble those obtained recently in primates ~Webb et al., 2005a!, and are consistent with an entirely cortical

7 origin of surround suppression only if the neurons providing suppression respond to drift rates somewhat higher than the majority of V1 neurons. The results are also consistent with a mixed source of suppressive signals, which might partly originate in V1 and partly be inherited from LGN. We performed a similar analysis ~not shown! for the low cutoff drift rate, the lowest mask drift rate at which suppression is half of its maximum. The results of this analysis were hardly informative: Stimuli drifting slowly tend to elicit sizeable responses both in LGN and in V1, and to cause both cross-orientation suppression and surround suppression. We did not observe a difference between the distributions of low cutoff drift rates for cross-orientation; surround suppression, V1 responses and LGN responses. Finally, we asked whether the temporal preferences of suppression exhibited by a V1 cell are related to the cell’s preferences for test stimuli. We compared on a cell-by-cell basis the preferred and cutoff drift rates for responses to test gratings with the preferred and cutoff drift rates for suppression by mask gratings ~Fig. 6!. The results of this analysis provide another distinction between crossorientation suppression and surround suppression. The temporal preferences of cross-orientation suppression correlate with those of the V1 neuron in which it is measured. We found a mild correlation ~R ⫽ 0.19 6 0.18, s.d., bootstrap estimate! between the preferred drift rate for responses to test gratings and the preferred drift rate for cross-orientation suppression by mask gratings ~Fig. 6A!. We discovered a stronger and more significant correlation ~R ⫽ 0.45 6 0.11! between cutoff drift rates ~Fig. 6C!. For a given cell, thus, the mask drift rates eliciting crossorientation suppression can to some extent be predicted from the test drift rates evoking responses. For surround suppression, instead, we found no clear correlation between preferred and cutoff drift rates of suppression and those of responses to test gratings. The correlation between preferred drift rates is at chance levels, R ⫽ 0.03 6 0.13 ~Fig. 6B!. The correlation between cutoff drift rates is only slightly higher and still not significant, R ⫽ 0.07 6 0.16 ~Fig. 6D!. As a result, the temporal properties of surround suppression received by a given cell seem independent of those of the same cell’s response. To summarize, the temporal preferences of individual V1 neurons are related to those of the signals mediating cross-orientation suppression, but not to the signals mediating surround suppression. This result reinforces the view that signals mediating the two form of suppression have different sources. For cross-orientation suppression, but not for surround suppression, it is likely that the signals causing suppression correspond to those that drive the responses ~consistent with an origin in LGN signals!. Suppression following cortical adaptation To measure the susceptibility to contrast adaptation of suppression, we recorded from 44 V1 neurons. We report here on 29 of these cells ~19 simple and 10 complex!, which fulfilled the following requirements: ~1! the maximum average response was greater than 5 spikes0s ~four cells excluded!; ~2! the cells were tuned for orientation ~three cells excluded! and ~3! the model used to summarize the data accounted for ⬎70% of the variance in the responses ~one cell excluded!. Finally, those cells whose index of suppression was less than 0.1 were not considered in the analysis of the model parameter values ~seven cells excluded!. The effects of suppression can be observed quantitatively by plotting the responses as a function of disk contrast, for different annulus contrasts ~Fig. 7A!. The responses grow sigmoidally with

8

S. Durand et al.

Fig. 6. Comparison of temporal properties of responses to test gratings and of suppression by mask gratings for individual V1 neurons. ~A! Preferred drift rate for response to test gratings against preferred mask drift rate for cross-orientation suppression R ⫽ 0.196 0.18 ~n ⫽ 33!. ~B! Same, for surround suppression, R ⫽ 0.03 6 0.13 ~n ⫽ 39!. ~C! Cutoff drift rate for response to test gratings against cutoff mask drift rate for cross-orientation suppression, R ⫽ 0.45 6 0.11 ~n ⫽ 33!. ~D! Same, for surround suppression, R ⫽ 0.07 6 0.16 ~n ⫽ 39!. Lines in A and C indicate linear regression ~with correlation R as indicated!.

disk contrast, and decrease with annulus contrast. This decrease corresponds mostly to a rightward shift along the contrast axis, illustrating a divisive effect ~DeAngelis et al., 1994!, as seen previously. Similar results can be observed for two other neurons ~Figs. 7E, 7G!. To consider the effects of surround suppression across different neurons, we fitted our data with the model used in the previous experiments. This model expresses responses as a function of test contrast ctest and mask contrast cmask . For experiments measuring surround suppression, test and mask are disk and annulus. To measure the strength of suppression in this model, we used a suppression index ~see Materials methods!. The model captures the effects of increasing annulus contrast ~Fig. 7!. It explains 95% of the variance for the data in Fig. 7A, and 87% and 96% of the variance for the data in Figs. 7E and 7G. Indeed, divisive models of this kind are known to do a good job at summarizing the responses to disk and annulus stimuli, as we have seen in Fig. 3 and has been demonstrated in other preparations ~Sceniak et al., 2001; Cavanaugh et al., 2002a!. We now turn to the effects of adaptation to the surrounding mask. Responses to disks alone and with 50% contrast annulus seen in Figs. 7A and 7B are replotted in Figs. 7C and 7D for clarity. For the cell in Fig. 7C, adaptation to the mask seemed to increase the response to the test alone ~Fig. 7D, open circles!. Indeed, while the maximal response R max was largely unchanged ~from 43 to 40

spikes0s!, the half-maximal contrast c50 decreased from 24% to 13%. An increase in responsiveness has been reported to follow adaptation of the surround in conditions that are similar to ours ~Pettet & Gilbert, 1992; DeAngelis et al., 1995!. We observed it in a few other cells. Adaptation also increased the responses measured in the presence of the mask, whose suppressive effect was decreased. Indeed, suppression prior to adaptation ~Fig. 7D, dashed curves, suppression index k ⫽ 10.3! appears stronger than following adaptation ~Fig. 7D, continuous curves, k ⫽ 2.0!. Similar effects can be observed for the cell in Figs. 7E and 7F, where adaptation reduced the suppression index k by 70%, and for the cell in Figs. 7G, 7H, where adaptation reduced k by 61%. The decrease in strength of suppression following adaptation to the annulus was observed in most of our cells. Adaptation markedly decreased the index of suppression k in 22029 cells ~Fig. 8A!. The strong tendency for the points to lie below the diagonal line indicates a consistent decrease in the strength of suppression. Error bars indicate 61 standard deviation estimated using the bootstrap method. This decrease is reflected in the histogram of the ratio of the parameter k prior to and following adaptation ~Fig. 8A, insert!. For most cells, this ratio is below one. On average, the suppression index k decreased from 2.49 6 0.55 ~s.e.! in the control measurement to 1.33 6 0.30 following adaptation to the annulus. These results are consistent with observations in primates, where pro-

Surround suppression

9

Fig. 7. Effects of adaptation on surround suppression in three example neurons. ~A! Mean firing rate responses of unit 19.1.14, 0.4 cycles0deg, disk 48, annulus 48-208, as a function of disk contrast, for five annulus contrasts between 0 and 50%. Responses were obtained in the control condition, following adaptation to the blank screen. Error bars are two standard errors of the mean ~N ⫽ 3!. Curves are fits of the model described in Methods. ~B! Responses obtained following adaptation to the annulus. ~C–D! Same unit as in A and B, we only replot the responses to disks alone and to disks with 50% contrast annulus for clarity. In D, the full line is the response after adaptation to an annulus. To facilitate comparison, we replotted the response after adaptation to the blank control screen form C ~dashed line!. ~E–F! Results from a second neuron. Unit 10.2.4, 1.2 cycles0deg, disk 2.48, annulus 3.08–8.08 (H–G! Results from a third neuron. Unit 19.4.4, 0.8 cycles0deg, disk 2.08, annulus 2.18–12.08. Dashed curves are replotted from A.

longed stimulation in the surrounding region of the receptive field was shown to reduce suppression originating from this region ~Cavanaugh et al., 2002b; Webb et al., 2005a!. By contrast, adaptation did not have a significant effect on the parameters describing the responses to the test alone. Maximal response Rmax was not substantially affected, with a mean value of 36.0 6 7.7 spikes0s in the control condition and 29.3 6 3.4 spikes0s following adaptation to the annulus ~Fig. 8B!. Halfmaximal contrast c50 was slightly decreased, from 21.1% 6 2.3% to 19.3% 6 1.6% ~Fig. 8C!, indicating a small increase in responsiveness. None of these effects are significant. Indeed the distributions of the ratios for these parameters ~Figs. 8B, 8C, inserts! are centered on 1, indicating an overall immunity to adaptation. These results confirm that the reduction in suppression index k was the main effect of adaptation. The marked effects of adaptation on the strength of surround suppression ~replotted in Fig. 9A! should be contrasted with the immunity to adaptation that we demonstrated for cross-orientation suppression ~Freeman et al., 2002! which is replotted in Fig. 9B. In that study we followed the same adaptation protocol described in the present study, only replacing the mask annulus with an orthogonal mask superimposed on the test. We defined an index of cross-orientation suppression, analogous to the index of surround suppression used here, and found this index to be unaffected by adaptation to the mask ~Fig. 9B!. A bootstrap analysis ~Efron & Tibshirani, 1993! of the distributions observed in Figs. 9A and 9B confirms that they are significantly different ~ p ⫽ 0.002!. Crossorientation suppression is thus immune to selective adaptation of those cortical neurons that respond to the mask, suggesting it does not rely on intracortical inhibition. Surround suppression, instead, does show on average a reduction following adaptation ~Fig. 9A!, indicating that its origins are at least partially cortical.

Discussion We measured the temporal preferences of signals underlying crossorientation and surround suppression. Cross-orientation suppression can be obtained with mask gratings drifting rather rapidly, whereas surround suppression has lower temporal resolution. We then compared the temporal properties of suppression to those of V1 and LGN responses. The temporal properties of cross-orientation suppression resemble those of LGN neurons, whereas the temporal properties of surround suppression are intermediate between those of LGN neurons and V1 neurons. We have found as well that the strength of surround suppression is reduced following adaptation to the annulus. The strength of cross-orientation suppression, instead, is unaffected by adaptation to the mask stimulus. Again, therefore, the properties of cross-orientation suppression resemble more those of LGN responses than those of V1 responses. Indeed, cat LGN neurons show little adaptation to stimuli drifting at the temporal frequencies that we tested ~Shou et al., 1996; Solomon et al., 2004! whereas V1 neurons are highly adaptable to these low frequencies. Conversely, the properties of surround suppression are more similar to those of V1 neurons. Our observation that surround suppression is observed preferentially with masks with drift rates similar or higher than those preferred by V1 neurons is consistent with an earlier report ~Li & Li, 1994!. These authors noted in passing a similarity in preferences for drift rate between signals mediating surround suppression and signals determining the cell’s response. Further, they showed that the bandwidth of suppression was larger than the bandwidth of responses. We confirmed and extended these results by considering a larger population, and by measuring the effects of surround suppression at different test contrasts. As with crossorientation suppression, these measurements confirm that suppres-

10

S. Durand et al.

Fig. 9. Comparison of surround suppression and cross-orientation suppression. ~A! Change in the index of surround suppression following adaptation to the mask annulus, replotted from the inset in A. Abscissa is the ratio of the index measured following adaptation to a blank stimulus and following adaptation to the mask annulus ~29 cells!. ~B! Results replotted from a similar study that probed the sources of cross-orientation suppression ~Freeman et al., 2002!. The cells that contribute to the distributions in A and B were generally different. Experimental design and data analysis were similar to the present study, with the annulus replaced by a mask placed over the center of the receptive field with an orientation orthogonal to the preferred. Abscissa reports the change in index of cross-orientation suppression following adaptation to the mask ~29 cells!. The distribution peaks at 1, which indicates no change.

Fig. 8. Summary of the effects of adaptation on all cells in our sample. Panels illustrate three key parameters of the model of surround suppression, measured following adaptation to the mask annulus ~ordinate! and in the control condition ~abscissa!. Points that lie above or below the diagonal indicate an increase or a decrease in the parameter value following adaptation to the mask annulus. Error bars indicate 61 standard deviation estimated using the bootstrap method. Closed square, circle, and diamond indicate the three cells in inserts indicate the ratio of the parameter measured with adaptation to the surround and the parameter measured in the control condition ~A! Values for the suppression index k. ~B! Values for the maximal response Rmax . ~C! Values for the half maximal contrast c50 .

sion has not a subtractive effect but a divisive effect. It allow us thus to measure suppression at its strongest, at low test contrasts. The outcome of our study is also in good agreement with recent results obtained in primates, where the surround suppression is readily seen with stimuli drifting at high frequencies ~Webb et al. 2005a! and the strength of surround suppression is reduced after prolonged exposure to a surrounding stimulus ~Cavanaugh et al., 2002b; Webb et al., 2005a!. For cross-orientation suppression, our results contribute to a mounting challenge of the intracortical

hypothesis, and suggest that the source of suppression might lie in signals from LGN rather than from cortex. This view is in good agreement with the lack of orientation selectivity seen in signals causing cross-orientation suppression ~DeAngelis et al., 1992!, and with the rapid onset of cross-orientation suppression, which seems to have a latency as fast as the responses themselves ~Smith et al., 2006!. Indeed, a feedforward model of cross-orientation suppression based on LGN responses forms the basis of current views of this phenomenon ~Carandini et al., 2002; Freeman et al., 2002; Li et al., 2006; Priebe & Ferster, 2006!. While cross-orientation suppression is mostly monocular, some suppression can also be seen dichoptically ~Li et al., 2005; Sengpiel & Vorobyov, 2005!. This dichoptic cross-orientation suppression could in principle operate both at the level of the LGN ~which is inhibited by binocular signals from the thalamic reticular formation! or at the level of V1 ~where responses are frankly binocular!. The bulk of the evidence suggests the second explanation ~Sengpiel & Vorobyov, 2005!. Dichoptic cross-orientation suppression could be mediated by inhibitory circuitry within the visual cortex. For surround suppression, instead, our results would be better explained by a combination of mechanisms involving LGN inputs and cortical inhibition, perhaps as along the lines proposed by Webb

Surround suppression et al. ~2005a!. Surround suppression might also be explained entirely by intracortical inhibition, but only if the relevant inhibitory neurons respond to higher drift rates than the majority of V1 neurons. Surround suppression is selective for orientation ~Blakemore & Tobin, 1972; DeAngelis et al., 1994; Sengpiel et al., 1997! and its onset is relatively slow ~Bair et al., 2003; Smith et al., 2006!. Surround suppression could be provided by a subclass of V1 neurons that respond to faster drift rates than the majority of cortical neurons. These might be inhibitory interneurons in V1. In the somatosensory system interneurons are typical in their ability to follow high stimulus frequencies ~Simons, 1978; Swadlow, 1988, 1989!. However, the visual preferences of interneurons in V1 are only beginning to be investigated ~Azouz et al., 1997; Palmer & Contreras, 2001!; we don’t know if they can respond to fast stimuli and thus cannot yet designate these neurons as responsible for surround suppression. A contribution to surround suppression could also originate in feedback from higher cortical areas ~Angelucci et al., 2002; Cavanaugh et al., 2002a; Levitt & Lund, 2002; Angelucci & Bullier, 2003; Angelucci & Bressloff, 2006; Smith et al., 2006!. Surround suppression seems to originate from a mechanism that integrates over regions larger than the classical receptive field. Indeed, being anatomically limited to the classical receptive field, horizontal connections within V1 seem unsuited to account for surround suppression ~Angelucci et al., 2002; Levitt & Lund, 2002; Angelucci & Bullier, 2003; Angelucci & Bressloff, 2006!. In this context, the fact that temporal resolution of surround suppression is somewhat higher than that of responses in area 17 could be explained by the fact that neurons in area 18 respond to slightly higher temporal frequencies than neurons in area 17 ~Movshon et al., 1978!. Moreover, in macaque extrastriate cortex, neurons in areas such as V3 ~Gegenfurtner et al., 1997! or MT ~Newsome et al., 1983! respond to stimuli somewhat faster than those that elicit responses in the majority of neurons in area V1; this difference perhaps arises thanks to direct geniculate input to these areas ~Sincich et al., 2004!. Finally, surround suppression in area V1 is likely to be partially due to simple inheritance of response properties of LGN neurons. Indeed, LGN cells do exhibit a form of surround suppression ~Walker et al., 1999; Jones et al., 2000; Solomon et al., 2002; Ozeki et al., 2004; Bonin et al., 2005!. Moreover, Ozeki et al. ~2004! showed no change in surround suppression after microinjection of GABA antagonist in V1, thus challenging any implication of inhibitory neurons in building surround suppression. Nonetheless, suppression seen in LGN does not explain all properties of surround suppression present in V1, in particular its orientation selectivity and susceptibility to adaptation. In summary, our findings corroborate the view that different suppressive mechanisms co-exist within visual cortex ~e.g., Sengpiel et al., 1998!. Our results confirm a distinction between the suppressive effects mediated near the receptive field center and those evoked from the surrounding areas. This view seems also true in humans ~Petrov et al., 2005! where these phenomena were studied perceptually and found to have markedly different properties.

Acknowledgments We thank Daniel Kiper, Valerio Mante, and Vincent Bonin for help with the experiments, and Preeti Verghese for helpful comments. Supported by the Swiss National Science Foundation and by the Human Frontiers Science Project Organization.

11 References Albrecht, D.G., Farrar, S.B. & Hamilton, D.B. ~1984!. Spatial contrast adaptation characteristics of neurones recorded in the cat’s visual cortex. Journal of Physiology (London) 347, 713–739. Albrecht, D.G. & Hamilton, D.B. ~1982!. Striate cortex of monkey and cat: Contrast response function. Journal of Neurophysiology 48, 217–237. Allison, J.D., Smith, K.R. & Bonds, A.B. ~2001!. Temporal-frequency tuning of cross-orientation suppression in the cat striate cortex. Visual Neuroscience 18, 941–948. Angelucci, A. & Bressloff, P.C. ~2006!. Contribution of feedforward, lateral and feedback connections to the classical receptive field center and extra-classical receptive field surround of primate V1 neurons. Progress in Brain Research 154, 93–120. Angelucci, A. & Bullier, J. ~2003!. Reaching beyond the classical receptive field of V1 neurons: Horizontal or feedback axons? Journal of Physiology (Paris) 97, 141–154. Angelucci, A., Levitt, J.B., Walton, E.J., Hupe, J.M., Bullier, J. & Lund, J.S. ~2002!. Circuits for local and global signal integration in primary visual cortex. Journal of Neuroscience 22, 8633–8646. Azouz, R., Gray, C.M., Nowak, L.G. & McCormick, D.A. ~1997!. Physiological properties of inhibitory interneurons in cat striate cortex. Cerebral Cortex 7, 534–545. Bair, W., Cavanaugh, J.R. & Movshon, J.A. ~2003!. Time course and time-distance relationships for surround suppression in macaque V1 neurons. Journal of Neuroscience 23, 7690–7701. Blakemore, C. & Tobin, E.A. ~1972!. Lateral inhibition between orientation detectors in the cat’s visual cortex. Experimental Brain Research 15, 439– 440. Bonds, A.B. ~1989!. Role of inhibition in the specification of orientation selectivity of cells in the cat striate cortex. Visual Neuroscience 2, 41–55. Bonin, V., Mante, V. & Carandini, M. ~2005!. The suppressive field of neurons in lateral geniculate nucleus. Journal of Neuroscience 25, 10844–10856. Brainard, D.H. ~1997!. The Psychophysics Toolbox. Spatial Vision 10, 433– 436. Carandini, M. ~2004!. Receptive fields and suppressive fields in the early visual system. In The Cognitive Neurosciences III. ed. Gazzaniga, M.S., pp. 313–326. Cambridge, MA: MIT Press. Carandini, M. & Ferster, D. ~1997!. A tonic hyperpolarization underlying contrast adaptation in cat visual cortex. Science 276, 949–952. Carandini, M., Heeger, D.J. & Movshon, J.A. ~1997!. Linearity and normalization in simple cells of the macaque primary visual cortex. Journal of Neuroscience 17, 8621–8644. Carandini, M., Heeger, D.J. & Senn, W. ~2002!. A synaptic explanation of suppression in visual cortex. Journal of Neuroscience 22, 10053–10065. Cavanaugh, J.R., Bair, W. & Movshon, J.A. ~2002a!. Nature and interaction of signals from the receptive field center and surround in macaque V1 neurons. Journal of Neurophysiology 88, 2530–2546. Cavanaugh, J.R., Bair, W. & Movshon, J.A. ~2002b!. Selectivity and spatial distribution of signals from the receptive field surround in macaque V1 neurons. Journal of Neurophysiology 88, 2547–2556. Chen, C.C., Kasamatsu, T., Polat, U. & Norcia, A.M. ~2001!. Contrast response characteristics of long-range lateral interactions in cat striate cortex. Neuroreport 12, 655– 661. DeAngelis, G.C., Anzai, A., Ohzawa, I. & Freeman, R.D. ~1995!. Receptive field structure in the visual cortex: Does selective stimulation induce plasticity? Proceedings of the National Academy of Sciences 92, 9682–9686. DeAngelis, G.C., Freeman, R.D. & Ohzawa, I. ~1994!. Length and width tuning of neurons in the cat’s primary visual cortex. Journal of Neurophysiology 71, 347–374. DeAngelis, G.C., Robson, J.G., Ohzawa, I. & Freeman, R.D. ~1992!. The organization of suppression in receptive fields of neurons in cat visual cortex. Journal of Neurophysiology 68, 144–163. Durand, S., Freeman, T.C.B., Mante, V., Kiper, D. & Carandini, M. ~2001!. Cross-orientation suppression in cat V1 with very fast stimuli. Society for Neuroscience Abstracts Vol. 27, Program No. 12.10. Durand, S., Mante, V., Freeman, T.C.B. & Carandini, M. ~2002!. Temporal properties of surround suppression in primary visual cortex. European Journal of Neuroscience, Abstract # 051.6. Efron, B. & Tibshirani, R.J. ~1993!. An introduction to the Bootstrap, vol. 57. New York: Chapman & Hall.

12 Fitzpatrick, D. ~2000!. Seeing beyond the receptive field in primary visual cortex. Current Opinion in Neurobiology 10, 438– 443. Freeman, T.C., Durand, S., Kiper, D.C. & Carandini, M. ~2002!. Suppression without inhibition in visual cortex. Neuron 35, 759–771. Gegenfurtner, K.R., Kiper, D.C. & Levitt, J.B. ~1997!. Functional properties of neurons in macaque area V3. Journal of Neurophysiology 77, 1906–1923. Heeger, D.J. ~1992!. Normalization of cell responses in cat striate cortex. Visual Neuroscience 9, 181–197. Hubel, D. & Wiesel, T. ~1965!. Receptive field and functional architecture in two nonstriate visual areas ~18–19! of the cat. Journal of Neurophysiology 28, 229–289. Ikeda, H. & Wright, M.J. ~1975!. Spatial and temporal properties of ‘sustained’ and ‘transient’ neurones in area 17 of the cat’s visual cortex. Experimental Brain Research 22, 363–383. Jagadeesh, B. & Ferster, D. ~1990!. Receptive field lengths in cat striate cortex can increase with decreasing stimulus contrast. Society for Neuroscience Abstracts 16, 293. Jones, H.E., Andolina, I.M., Oakely, N.M., Murphy, P.C. & Sillito, A.M. ~2000!. Spatial summation in lateral geniculate nucleus and visual cortex. Experimental Brain Research 135, 279–284. Kapadia, M.K., Westheimer, G. & Gilbert, C.D. ~1999!. Dynamics of spatial summation in primary visual cortex of alert monkeys. Proceedings of the National Academy of Sciences 96, 12073–12078. Lehmkuhle, S., Kratz, K.E., Mangel, S.C. & Sherman, S.M. ~1980!. Spatial and temporal sensitivity of X- and Y-cells in dorsal lateral geniculate nucleus of the cat. Journal of Neurophysiology 43, 520–541. Levitt, J.B. & Lund, J.S. ~2002!. The spatial extent over which neurons in macaque striate cortex pool visual signals. Visual Neuroscience 19, 439– 452. Li, B., Peterson, M.R., Thompson, J.K., Duong, T. & Freeman, R.D. ~2005!. Cross-orientation suppression: Monoptic and dichoptic mechanisms are different. Journal of Neurophysiology 94, 1645–1650. Li, B., Thompson, J.K., Duong, T., Peterson, M.R. & Freeman, R.D. ~2006!. Origins of cross-orientation suppression in the visual cortex. Journal of Neurophysiology 96, 1755–1764. Li, C. & Li, W. ~1994!. Extensive integration beyond the classical receptive field of cat’s striate cortical neurons—classification and tuning properties. Vision Research 34, 2337–2356. Maffei, L. & Fiorentini, A. ~1976!. The unresponsive regions of visual cortical receptive fields. Vision Research 13, 1255–1267. Maffei, L., Fiorentini, A. & Bisti, S. ~1973!. Neural correlate of perceptual adaptation to gratings. Science 182, 1036–1038. Morrone, M.C., Burr, D.C. & Maffei, L. ~1982!. Functional implications of cross-orientation inhibition of cortical visual cells. I. Neurophysiological evidence. Proceedings of the Royal Society of London, Series B 216, 335–354. Movshon, J.A., Thompson, I.D. & Tolhurst, D.J. ~1978!. Spatial and temporal contrast sensitivity of neurones in areas 17 and 18 of the cat’s visual cortex. Journal of Physiology (London) 283, 101–120. Newsome, W.T., Gizzi, M.S. & & Movshon, J.A. ~1983!. Spatial and temporal properties of neurons in macaque MT. Investigative Ophthalmology & Visual Science Supplement 24, 106. Ohzawa, I., Sclar, G. & Freeman, R.D. ~1985!. Contrast gain control in the cat visual system. Journal of Neurophysiology 54, 651– 665. Orban, G.A., Hoffman, K.-P. & Duysens, J. ~1985!. Velocity selectivity in the cat visual system.I. Responses of LGN cells to moving bar stimuli: A comparison with cortical areas 17 and 18. Journal of Neurophysiology 54, 1026–1049. Ozeki, H., Sadakane, O., Akasaki, T., Naito, T., Shimegi, S. & Sato, H. ~2004!. Relationship between excitation and inhibition underlying size tuning and contextual response modulation in the cat primary visual cortex. Journal of Neuroscience 24, 1428–1438. Palmer, L.A. & Contreras, D. ~2001!. Differential contrast sensitivity of excitatory and inhibitory neurons in cat area 17. Society for Neuroscience Abstracts 27, Program No.821.61. Pelli, D.G. ~1997!. The VideoToolbox software for visual psychophysics: Transforming numbers into movies. Spatial Vision 10, 437– 442.

S. Durand et al. Petrov, Y., Carandini, M. & McKee, S. ~2005!. Two distinct mechanisms of suppression in human vision. Journal of Neuroscience 25, 8704–8707. Pettet, M.W. & Gilbert, C.D. ~1992!. Dynamic changes in receptivefield size in cat primary visual cortex. Proceedings of the National Academy of Sciences 89, 8366–8370. Priebe, N.J. & Ferster, D. ~2006!. Mechanisms underlying crossorientation suppression in cat visual cortex. Nature Neuroscience 9, 552–561. Sanchez-Vives, M.V., Nowak, L.G. & McCormick, D.A. ~2000!. Membrane mechanisms underlying contrast adaptation in cat area 17 in vivo. Journal of Neuroscience 20, 4267– 4285. Saul, A.B. & Humphrey, A.L. ~1990!. Spatial and temporal response properties of lagged and nonlagged cells in cat lateral geniculate nucleus. Journal of Neurophysiology 64, 206–224. Saul, A.B. & Humphrey, A.L. ~1992!. Temporal-frequency tuning of direction selectivity in cat visual cortex. Visual Neuroscience 8, 365–372. Sceniak, M.P., Hawken, M.J. & Shapley, R. ~2001!. Visual spatial characterization of macaque V1 neurons. Journal of Neurophysiology 85, 1873–1887. Sceniak, M.P., Ringach, D.L., Hawken, M.J. & Shapley, R. ~1999!. Contrast’s effect on spatial summation by macaque V1 neurons. Nature Neuroscience 2, 733–739. Sengpiel, F., Baddeley, R.J., Freeman, T.C., Harrad, R. & Blakemore, C. ~1998!. Different mechanisms underlie three inhibitory phenomena in cat area 17. Vision Research 38, 2067–2080. Sengpiel, F., Sen, A. & Blakemore, C. ~1997!. Characteristics of surround inhibition in cat area 17. Experimental Brain Research 116, 216–228. Sengpiel, F. & Vorobyov, V. ~2005!. Intracortical origins of interocular suppression in the visual cortex. Journal of Neuroscience 25, 6394– 6400. Shou, T., Li, X., Zhou, Y. & Hu, B. ~1996!. Adaptation of visually evoked responses of relay cells in the dorsal lateral geniculate nucleus of the cat following prolonged exposure to drifting gratings. Visual Neuroscience 13, 605– 613. Simons, D. ~1978!. Response properties of vibrissa units in rat SI somatosensory neocortex. Journal of Neurophysiology 41, 798–820. Sincich, L.C., Park, K.F., Wohlgemuth, M.J. & Horton, J.C. ~2004!. Bypassing V1: A direct geniculate input to area MT. Nature Neuroscience 7, 1123–1128. Skottun, B.C., De Valois, R.L., Grosof, D.H., Movshon, J.A., Albrecht, D.G. & Bonds, A.B. ~1991!. Classifying simple and complex cells on the basis of response modulation. Vision Research 31, 1079–1086. Smith, M.A., Bair, W. & Movshon, J.A. ~2006!. Dynamics of suppression in macaque primary visual cortex. Journal of Neuroscience 26, 4826– 4834. Solomon, S.G., Peirce, J.W., Dhruv, N.T. & Lennie, P. ~2004!. Profound contrast adaptation early in the visual pathway. Neuron 42, 155–162. Solomon, S.G., White, A.J. & Martin, P.R. ~2002!. Extraclassical receptive field properties of parvocellular, magnocellular, and koniocellular cells in the primate lateral geniculate nucleus. Journal of Neuroscience 22, 338–349. Swadlow, H. ~1988!. Efferent neurons and suspected interneurons in binocular visual cortex of the awake rabbit: Receptive fields and binocular properties. Journal of Neurophysiology 59, 1162–1187. Swadlow, H. ~1989!. Efferent neurons and suspected interneurons in S-1 vibrissa cortex of the awake rabbit: Receptive fields and axonal properties. Journal of Neurophysiology 62(1), 288–308. Walker, G.A., Ohzawa, I. & Freeman, R.D. ~1999!. Asymmetric suppression outside the classical receptive field of the visual cortex. Journal of Neuroscience 19, 10536–10553. Webb, B.S., Dhruv, N.T., Solomon, S.G., Tailby, C. & Lennie, P. ~2005a!. Early and late mechanisms of surround suppression in striate cortex of macaque. Journal of Neuroscience 25, 11666–11675. Webb, B.S., Tinsley, C.J., Vincent, C.J. & Derrington, A.M. ~2005b!. Spatial distribution of suppressive signals outside the classical receptive field in lateral geniculate nucleus. Journal of Neurophysiology 94, 1789–1797.

Temporal properties of surround suppression in cat ... - Matteo Carandini

suppression with surround drift rates as high as 21 Hz. We then measured the susceptibility of suppression to .... intracortical hypothesis ~top! ascribes cross-orientation suppression to ..... It is of interest to follow the same approach for surround.

301KB Sizes 2 Downloads 219 Views

Recommend Documents

Temporal properties of surround suppression in cat ... - Matteo Carandini
To analyze these responses, we fit them with a descriptive model, a function of disk .... model was not describing the data in a satisfactory way. The strength of ...

Article - Matteo Carandini
May 21, 2008 - Functional models of the early visual system should predict responses not only ..... The model provided excellent fits, accounting for 79% of the.

NN0406 NV_ERR.indd - Matteo Carandini
in fog? They might think they are driving slowly. ... When fog reduces contrast, drivers may think .... The authors' approach also accounts for aspects of the data.

Article - Matteo Carandini
May 21, 2008 - this model of fast adaptation fits the data practically as well as the full set (Figure ..... in the responses to natural images: the same analysis on the response to ... Indeed, the predicted time courses are less tran- sient than the

Article - Matteo Carandini
May 21, 2008 - For the example cell (Figure 1B), the fraction of stimulus-driven variance in the .... ually (Figure 3B, compare black and red) and predict the re- ...... sensitivity regulation in primate outer retina: The horizontal cell network. J.

NN0406 NV_ERR.indd - Matteo Carandini
were able to test the Bayesian model much more thoroughly ... that an observer will judge a test stimulus ... so successful in the domain of speed per- ception ...

NN0406 NV_ERR.indd - Matteo Carandini
A Bayesian model of visual motion perception describes how the brain combines assumption with evidence. A new ... these two probability distributions to obtain.

NN0406 NV_ERR.indd - Matteo Carandini
Visual scenes contain a vari- ety of contrasts, including regions of low or even zero contrast4. At high contrast, neural circuits devoted to visual motion may have.

Diverse coupling of neurons to populations in ... - Matteo Carandini
Apr 6, 2015 - with this view, intracellularin vivomeasurements indicated that popu- ..... field of view of ,120u360u, extending in front and to the right of the ...

Nonlinear processing in LGN neurons - Matteo Carandini
convolution of the map of stimulus contrast S(x,t) with a receptive field F(x,t):. [ ]( ). () S F ..... Role of inhibition in the specification of orientation selectivity of cells.

Nonlinear processing in LGN neurons - Matteo Carandini
operate linearly (Cai et al., 1997; Dan et al., 1996). Their response L(t) is the convolution of the map of stimulus contrast S(x,t) with a receptive field F(x,t):. [ ]( ).

Mapping of Stimulus Energy in Primary Visual Cortex - Matteo Carandini
Mar 9, 2005 - cal imaging study of primary visual cortex (V1) by Basole, White, and .... The costs of publication of this article were defrayed in part by the ...

Integration of visual motion and locomotion in ... - Matteo Carandini
Nov 3, 2013 - separate data segment (the 'test set'), we defined a prediction quality ..... learn the stable mapping between movements and visual flow. In a.

Diverse coupling of neurons to populations in ... - Matteo Carandini
Apr 6, 2015 - V1 were bulk-loaded with Oregon Green BAPTA-1 dye and their ...... a, A recurrent network where excitatory cells (triangles) send synaptic.

Integration of visual motion and locomotion in ... - Matteo Carandini
Nov 3, 2013 - activity by 50% to 200% (Supplementary Fig. 6). ..... Previous data suggested that the effect of locomotion was binary6, as would be expected if ...

Adaptation to contingencies in macaque primary ... - Matteo Carandini
activity they receive, even when they do not initially ... in the attributes of the stimuli they receive. We ... and were vignetted by a square window of optimal size.

Adaptation to contingencies in macaque primary ... - Matteo Carandini
all types of contingency. 1. INTRODUCTION ... An alternative explanation is that cortical neurons ..... compound stimuli had higher contrast energy than the.

Five key factors determining pairwise correlations ... - Matteo Carandini
May 27, 2015 - disjoint observations in a vast literature on pairwise correlations and ..... C: average firing rate of neurons as a function of spike isolation. For d= ...... Truccolo W, Eden UT, Fellows MR, Donoghue JP, Brown EN. A point process ...

temporal response properties of local field potentials in ...
signals were recorded with a data acquisition system (Plexon Inc.,. Dallas, TX .... R cosð/Þ þ A where u and v are the anatomical distances from the rostral pole.

From circuits to behavior: a bridge too far? - Matteo Carandini
the hardware to the operation of the computer. What discovery would ... (a) The Center for. Neural Circuits and Behavior at University of. California, San Diego (photo by Bassam Atallah). (b) An exceedingly literal interpretation of this article's vi

Five key factors determining pairwise correlations ... - Matteo Carandini
May 27, 2015 - Submitted 29 January 2015; accepted in final form 22 May 2015. Schulz DP ... correlations on neuronal networks increases with the size of the population .... a CRT monitor (Sony Trinitron 500PS, refresh rate of 125 Hz, mean luminance o

Normalization as a canonical neural computation - Matteo Carandini
Nov 23, 2011 - intensity distributions. d | The same data as in part c plotted as a function of local contrast ... The responses of a V1 neuron to a test grating that drives responses are ..... divisive signals is generally hard to distinguish based

Monitoring of Temporal First-order Properties with ...
aggregations and grouping operations in our language mimics that of SQL. As ... We first compare the performance of our prototype implementation with the.

Spatial and temporal variability of seawater properties ...
open, sandy coastal area known for the occurrence of patches of fairly large amounts of muddy sediments ... winds from NE and from SW account for, respectively, $22% and ..... exhibits an alternating pattern of offshore (positive) and onshore.