3624 • The Journal of , April 5, 2006 • 26(14):3624–3625

Journal Club Surround Suppression in the Early

Matthew A. Smith Center for the Neural Basis of Cognition, Carnegie Mellon University, Pittsburgh, Pennsylvania 15232 Review of Webb et al. (http://www.jneurosci.org/cgi/content/full/25/50/11666)

Introduction jneurosci.org/cgi/content/full/25/50/ 11666/FIG7)]. Thus, surround suppression In their description of orientation-tuned 11666) exploited the different response was activated by stimuli that drive the CRF cells in primary visual cortex (V1) of the properties of subcortical and cortical of most V1 poorly or not at all, im- macaque, Hubel and Wiesel (1968) found neurons to distinguish between these plicating a mechanism that acts at or before that some neurons responded less when potential mechanisms for surround the input layers of V1. However, the orien- they lengthened an optimally oriented bar suppression. tation selectivity of surround suppression stimulus. They termed cells that exhibited The authors first identified the optimal instead points to a mechanism beyond the this behavior to be “hypercomplex”. Nu- grating stimulus for the CRF with respect V1 input layers. Webb et al. (2005) propose merous studies have since explored this to orientation, size, and spatial and tem- that two separate mechanisms underlie sur- phenomenon, now known commonly poral frequency. Then, using a CRF stim- round suppression, and designed two ex- as surround suppression. At least three ulus with these parameters, they varied periments to test this hypothesis. sources have been proposed (Fig. 1) to ac- the spatial frequency of the surround an- Neurons in V1 and the LGN display a count for this kind of suppression by nulus as presented both parallel and or- temporary loss of responsiveness after stimuli outside the classical thogonal to the CRF stimulus. The aver- prolonged exposure to a high- (CRF). Long-range lateral or horizontal age spatial frequency preferred by the stimulus. Webb et al. (2005) took advan- connections within V1 were originally surround was 2.2 times lower as well as tage of this “contrast adaptation” to test suggested to form the circuitry underlying much broader than that preferred by the the hypothesized dual mechanisms of sur- surround suppression (Gilbert and Wie- CRF. Furthermore, very high frequencies round suppression. They determined the sel, 1983). This was based on anatomical (five times the CRF peak) and a spatially susceptibility of the surround to adapta- evidence showing connections between uniform annulus elicited powerful sup- tion with gratings of different contrasts distant neurons of like orientation, and pression [Webb et al. (2005), their Fig. 3 and orientations as well as a spatially uni- physiological evidence that suppression is (http://www.jneurosci.org/cgi/content/ form field. Adaptation to both optimal strongest at the orientation that matches full/25/50/11666/FIG3)], although these and orthogonal gratings reduced the ef- the preferred receptive field stimulus. stimuli were poor at driving the CRF. fectiveness of the surround, producing a More recently, feedback from higher cor- Webb et al. (2005) found similar results fourfold increase in the contrast necessary tical areas has been proposed as a source when testing the temporal frequency se- to suppress responses with an optimal because of the relatively slow dynamics of lectivity of the surround. Although the grating. This is higher than would be pre- surround suppression and the lack of a CRF is sharply tuned for temporal fre- dicted based on the sensitivity of LGN strong dependence on cortical distance quency, the surround was broadly tuned neurons. However, adaptation to a spa- (Bair et al., 2003). Another suggestion is –nearly flat within the range they tested. tially uniform annulus produced no that activation of orientation-tuned sur- In addition, they were able to elicit sup- change in the effectiveness of the sur- round suppression in the lateral genicu- pression from the surround at very low round [Webb et al. (2005), their Fig. 9 late nucleus (LGN) leads to reduced exci- (0.5 Hz) and high (30 Hz) temporal fre- (http://www.jneurosci.org/cgi/content/ tatory drive to V1 (Ozeki et al., 2004). As quencies [Webb et al. (2005), their Fig. 5 full/25/50/11666/FIG9)], although this the study of surround suppression has (http://www.jneurosci.org/cgi/content/full/ stimulus is a potent one for LGN neurons. evolved, it has become increasingly diffi- 25/50/11666/FIG5)], outside the range to The strength of adaptation seems incon- cult to reconcile these three. In a recent which most V1 neurons respond. Finally, sistent with a thalamic source for sur- study published in The Journal of Neuro- they tested the contrast sensitivity of the sur- round suppression, but the similar adap- science, Webb et al. (2005; http://www. round to optimally oriented grating stimuli, tation to optimal and orthogonal gratings orthogonal gratings, and a spatially uniform is not expected from an orientation- field. As expected, the surround was most selective cortical mechanism. Received Jan. 18, 2006; revised, accepted Feb. 17, 2006. sensitive to an optimal grating stimulus. Visual signals from the two eyes are Correspondence should be addressed to Dr. Matthew A. Smith, Center Less predictable was their finding of signifi- segregated at the level of the LGN and the for the Neural Basis of Cognition, Carnegie Mellon University, 4400 Fifth cant sensitivity to a spatially uniform field, input layers of V1, and binocular interac- Avenue, Mellon Institute Room 115, Pittsburgh, PA 15213. E-mail: [email protected]. exceeding that to an orthogonal grating tions become significant only outside DOI:10.1523/JNEUROSCI.0236-06.2006 [Webb et al. (2005), their Fig. 7 (http:// these layers (Hubel and Wiesel, 1968). Copyright©2006SocietyforNeuroscience 0270-6474/06/263624-02$15.00/0 www.jneurosci.org/cgi/content/full/25/50/ Webb et al. (2005) used an interocular Smith • Journal Club J. Neurosci., April 5, 2006 • 26(14):3624–3625 • 3625

properties consistent with suppression from the LGN or V1 input layers would also be the ones with a fast time course of suppression. Furthermore, although Webb et al. (2005) identify two putative sources of surround suppression, the loci remain imprecise. The “early” mecha- nism might be located in either the LGN or V1 input layers, and the “late” mecha- nism might arise from horizontal connec- tions within the V1 or feedback from the extrastriate cortex. Finally, additional knowledge of the LGN suppressive field is necessary to determine what role it plays in surround suppression in the V1. Bonin et al. (2005) demonstrated that a contrast gain control model is able to explain di- Figure 1. Schematic of three proposed mechanisms for surround suppression. The open triangles represent excitatory neu- verse suppressive phenomenon seen in rons, and the filled circles represent inhibitory interneurons. The recorded is indicated with a triangle shaded gray. This LGN responses. Whether V1 surround neuron is driven by the CRF stimulus (shown as a small patch of grating) and suppressed in the presence of a surround stimulus (shown as a larger annular grating). A, In the horizontal connection model, V1 neurons with receptive fields stimulated by the suppression can be partially explained by surround stimulus suppress the recorded neuron. This is accomplished via projections to an inhibitory interneuron. B, A different expanding such a model remains an open model effects suppression through feedback from extrastriate cortex. Here, V1 inputs are pooled by an extrastriate neuron with a question. Webb et al. (2005) have pro- large receptive field. The recorded neuron is suppressed via feedback connections to an inhibitory interneuron. C, A third model vided an important conceptual frame- generatessurroundsuppressionwithintheLGN,notrequiringextensivecorticalcircuitry.Theillustratedmechanisminvolveslocal work as a starting point. Future research inhibitoryconnectionswithintheLGN,butrelatedmodels(datanotshown)mightbeconstructedbyinvokingretinalgaincontrol will be needed to guide our understanding or feedback from visual cortex. of surround suppression in the V1, as well as its link to other forms of contextual transfer experiment to test whether sur- mechanisms underlie surround suppres- modulation by stimuli outside the CRF. round suppression occurs before or after sion: one that arrives early and is consis- the signals from the two eyes intermingle. tent with an origin in the LGN or input References They delivered an optimal CRF stimulus layers of V1; the other arrives later and is Bair W, Cavanaugh JR, Movshon JA (2003) to the dominant eye and tested the degree compatible with horizontal V1 connec- Time-course and time-distance relationships of suppression when a surround grating tions or feedback from extrastriate cortex. for surround suppression in macaque V1 neu- rons. J Neurosci 23:7690–7701. was presented to each of the two eyes. Al- Their data are compelling evidence that a Bonin V, Mante V, Carandini M (2005) The though suppression was stronger when single mechanism is insufficient to ac- suppressive field of neurons in lateral genicu- the stimulus was delivered to the sur- count for the variety of properties ob- late nucleus. J Neurosci 23:10844–10856. round of the dominant eye, significant served in surround suppression. How- Gilbert CD, Wiesel TN (1983) Clustered intrin- suppression remained to a nondominant ever, important questions remain. Bair et sic connections in cat visual cortex. J Neurosci eye stimulus [Webb et al. (2005), their al. (2003) reported that surround sup- 3:1116–1133. Hubel D, Wiesel T (1968) Receptive fields and Fig. 10 (http://www.jneurosci.org/cgi/ pression in V1 neurons could be delayed Ͼ functional architecture of monkey striate cor- content/full/25/50/11666/FIG10)]. If sur- by 40 ms after response onset, but it tex. J Physiol (Lond) 195:215–243. round suppression were generated by a could also precede it by Ͼ20 ms. Two Ozeki H, Sadakane O, Akasaki T, Naito T (2004) mechanism located before the binocular mechanisms for surround suppression Relationship between excitation and inhibi- combination of signals, this result would could account for this wide range of de- tion underlying size tuning and contextual re- not be expected. Instead, the presence of lays. To reach that conclusion, it would be sponse modulation in the cat primary visual interocular transfer suggests that suppres- critical to examine the timing of surround cortex. J Neurosci 24:1428–1438. Webb BS, Dhruv NT, Solomon SG, Tailby C, Len- sion arises from a cortical mechanism be- suppression along with the experiments nie P (2005) Early and late mechanisms of yond the input layers of V1. performed by Webb et al. (2005). Presum- surround suppression in striate cortex of ma- The authors conclude that two distinct ably, neurons that are identified as having caque. J Neurosci 25:11666–11675.