pubmed.ncbi.nlm.nih.gov

The role of feedback in shaping the extra-classical receptive field of cortical neurons: a recurrent network model - PubMed

  • ️Sun Jan 01 2006

The role of feedback in shaping the extra-classical receptive field of cortical neurons: a recurrent network model

Lars Schwabe et al. J Neurosci. 2006.

Abstract

The responses of neurons in sensory cortices are affected by the spatial context within which stimuli are embedded. In the primary visual cortex (V1), orientation-selective responses to stimuli in the receptive field (RF) center are suppressed by similarly oriented stimuli in the RF surround. Surround suppression, a likely neural correlate of perceptual figure-ground segregation, is traditionally thought to be generated within V1 by long-range horizontal connections. Recently however, it has been shown that these connections are too short and too slow to mediate fast suppression from distant regions of the RF surround. We use an anatomically and physiologically constrained recurrent network model of macaque V1 to show how interareal feedback connections, which are faster and longer-range than horizontal connections, can generate "far" surround suppression. We provide a novel solution to the puzzle of how surround suppression can arise from excitatory feedback axons contacting predominantly excitatory neurons in V1. The basic mechanism involves divergent feedback connections from the far surround targeting pyramidal neurons sending monosynaptic horizontal connections to excitatory and inhibitory neurons in the RF center. One of several predictions of our model is that the "suppressive far surround" is not always suppressive, but can facilitate the response of the RF center, depending on the amount of excitatory drive to the local inhibitors. Our model provides a general mechanism of how top-down feedback signals directly contribute to generating cortical neuron responses to simple sensory stimuli.

PubMed Disclaimer

Figures

Figure 1.
Figure 1.

Different components of the RF center and surround of V1 neurons and their hypothesized anatomical substrates. In the center of the figure is a schematic diagram of the various components of the RF center and surround of an example V1 neuron; these are surrounded by four identical size-tuning curves for this neuron, over which an icon and arrow indicate the size of the respective RF component. Black and gray curves are size-tuning curves measured at high- and low-stimulus contrast, respectively. The white square area is the mRF or RF center; this is the RF region over which presentation of optimally oriented stimuli evokes spikes from the cell. Experimentally, the mRF is measured using small, high-contrast moving stimuli and delimiting the area where spikes are evoked from the cell (icon on the top-left size-tuning curve) (Hubel and Wiesel, 1962; Barlow et al., 1967); the arrow under the mRF icon indicates the size of the mRF relative to the peak of the high-contrast size-tuning curve. The dashed ring is the hsRF; this is measured by presenting high-contrast gratings of increasing radius (icon on the center-left size-tuning curve) and defined as the stimulus radius at the peak response of the neuron (arrow under the hsRF icon). The region between the mRF and the hsRF is the region over which presentation of high-contrast gratings at the same orientation as the center grating facilitates the response of the cell to optimally oriented gratings in the center. The hsRF radius is about twice that of the mRF. The continuous ring is the lsRF; this is measured and defined as the hsRF size, but using low-contrast gratings of increasing radius (icon on the center-right size-tuning curve). The region between the hsRF and the lsRF is the region over which presentation of gratings at the same orientation as the center grating suppresses or facilitates the response of the cell to optimally oriented gratings in the center, depending on the contrast of the grating. Note the shift to the right of the peak response at low contrast (Sceniak et al., 1999). Gray area, RF surround. We consider two separate regions of the surround depending on their proximity to the RF center: (1) the near surround is the region between the mRF and the lsRF, (2) the far surround is the region outside the lsRF over which presentation of stimuli at the same orientation as the center stimulus usually suppresses the response of the cell to optimally oriented gratings in the center. Intra-areal V1 horizontal connections (red) are commensurate with the lsRF size of their V1 neurons of origin, whereas extrastriate FB (blue) connections to V1 are commensurate with the full spatial scale of the center and surround field of V1 neurons (Angelucci et al., 2002b). Feedforward geniculocortical connections (data not shown) are commensurate with the hsRF size of their recipient V1 neurons (Angelucci and Sainsbury, 2006).

Figure 2.
Figure 2.

Basic architecture of the recurrent network model. Schematic diagram of the connections used in the network model. Different connection types are indicated as color-coded arrows. Purple and black boxes represent populations of excitatory (E) or inhibitory (I) V1 neurons, respectively, labeled according to the position of their RF center relative to that of the center neurons. Accordingly, ctr are the neurons in the RF center or mRF; nr are the neurons in the near surround, of which those closer to the center fall within the hsRF of the center neurons, whereas those located farther from the center fall within the lsRF of the center neurons; far are the neurons in the far surround. EFF, Excitatory neurons in other V1 layers sending feedforward afferents to the E neurons in V1 layers 2/3. EFB, excitatory neurons in extrastriate cortex sending feedback projections to the E neurons in V1. FB connections are spatially highly divergent and convergent. Note the absence of direct FB inputs to I neurons. The latter receive monosynaptic inputs only from V1 horizontal connections (red arrows) and from local E neurons via local recurrent connections (purple arrows). Icons at the bottom represent the different components of the RF center and surround (same conventions as in Fig. 1), with red areas indicating RF components that are stimulated when each respective submodule is consecutively (from left to right) activated by a stimulus of increasing radius.

Figure 3.
Figure 3.

Functional specification of the network model. a, Different spatial scales used in the model for the FB (black) and the lateral (or horizontal; gray) connections. Plotted is the normalized strength of lateral and FB connections to a postsynaptic neuron as a function of the distance of the RF centers of the presynaptic neurons from the RF center of the postsynaptic neuron. Note the different scales on the x axes (the top one applies to the FB connections, the bottom to the lateral connections). b, Firing rate of the V1 local excitatory (E; gray) and inhibitory (I; black) neurons in the model, plotted against the input current. c, Input current of the V1 excitatory neuron as a function of stimulus contrast.

Figure 4.
Figure 4.

Contrast-dependent spatial summation in the network model. a, Response (spikes/second) of the center excitatory (Ectr; purple) and inhibitory (Ictr; black) neurons as a function of the size of the afferent stimulus (in degrees) for two different stimulus contrasts (85 and 15%, solid and dashed lines, respectively). Icons at the top represent the different components of the RF center and surround (as in Figs. 1, 2), with red areas indicating the RF regions that are activated at the indicated points in the size-tuning curve. These same icons are shown in d–f to indicate the respective active network components and pathways. b, c, Input currents to the center excitatory neurons (b) and to the center inhibitory neurons (c) as a function of the afferent stimulus size (in degrees). The different input types are color coded according to the legend in f. Icons under the abscissa in c show the stimulus used and are valid for a–c. d–f, A sequence of diagrams highlighting, for increasing stimulus size, the network components that are active and the major afferent pathways from the active regions to the center neurons. Conventions are as in Figure 2.

Figure 5.
Figure 5.

Contrast-dependent suppression and facilitation from the far surround, in the network model. a, Response (spikes/second) of the center excitatory (Ectr; purple) and inhibitory (Ictr; black) neurons to a high-contrast (85%) central stimulus of 0.5° radius (i.e., just larger than the hsRF of the neuron; Fig. 4a) plotted against the inner radius of a high-contrast (85%) annular stimulus of an 8° outer radius presented together with the central stimulus (stimulus configuration shown in b). Other conventions are as in Figure 4. b, Same as in a, but for excitatory neurons in the near surround (Enr). Icon under the abscissa represents the stimulus used in a and b. c, Same as in a but the configuration of the stimulus used (shown in d) consists of a low-contrast (15%) central stimulus (0.5° radius) presented together with a high-contrast (85%) annular stimulus in the far surround. d, Same as in c, but for excitatory neurons in the near surround (Enr). Icon under the abscissa represents the stimulus used in c and d. e, Diagram highlighting, for the RF regions activated by the stimulus (red areas in the icon at the bottom), the network components that are active and the major afferent pathways from the active regions to the center neurons. Conventions are as in Figures 2 and 4d–f.

Figure 6.
Figure 6.

Population activity profile of the excitatory neurons across V1. a, Activity profile for a small (0.4° radius; thick line) and a large (1.5° radius; thin line) high-contrast (85%) stimulus. Plotted are the firing rates (spikes/second) of the excitatory neurons against their location in V1. b, Activity profile for a small (0.5° radius) high-contrast (85%) central stimulus presented alone (thick line) or together (thin line) with a high-contrast (85%) annular stimulus (3.2 and 8° inner and outer radius, respectively). The cortical location of the neurons were computed by converting the distance in degrees between their RF centers into cortical distance, using a cortical magnification factor of 2.3 mm/° (see Materials and Methods).

Figure 7.
Figure 7.

Center-surround interactions with and without feedback, in the network model. a, Response (spikes/second) of the center excitatory neurons (Ectr) to the central stimulus alone (small grating; 0.68° radius), and to the central stimulus presented together with a surround stimulus (large grating; surround stimulus of 0.68° inner radius and 8° outer radius) in the presence of FB (gray bars) or with the FB connections inactivated (white bars). The center plus surround stimulus was presented at low saliency (38 vs 15% contrast; see Results for definition of saliency). b, same as in a but for a high-saliency center plus surround stimulus (77 vs 15% contrast). The size of the central stimulus was always fitted to the stimulus size at the peak response of the size-tuning curve (i.e., the optimal size), which varies with the stimulus contrast; thus, in b, the central stimulus had a radius of 0.49°, which is the optimal stimulus size at 77% contrast. c, Response change (see Results for definition) of the center excitatory neurons as a function of stimulus saliency, with and without FB. For each contrast, the central stimulus was presented at the optimal size.

Figure 8.
Figure 8.

Dynamics and latency of surround suppression in the network model. a, Response (spikes/second) of the center excitatory neurons (Ectr) to a high-contrast (85%) central stimulus (0.5° radius) first presented alone and, after 200 ms, together with a high-contrast annular stimulus in the far surround of varying inner radius (2.5–7°), plotted against the time from the onset of the central stimulus. In a–c, the arrow points in the direction of the decreasing inner radius of the annulus, as indicated by the stimulus icon. b, Suppression strength of the response of the center neurons induced by a stimulus in the far surround, plotted against the time from the onset of the suppression. Suppression strength is defined as 100*(RC − RCS/RC), where RC and RCS are the response of the neurons to the presentation of the central stimulus alone and of the central stimulus shown together with the surround stimulus, respectively. c, Responses (spikes/second) of inhibitory neurons in the center (black) and excitatory neurons in the near surround (gray) for annuli of varying inner radius. d, Suppression strength of the steady-state response of the center excitatory neurons as a function of the inner radius of the surround annulus. e, Latency of suppression onset (defined as the time from the onset of the annulus at which suppression strength reached 3%) as a function of the inner radius of the surround annulus.

Similar articles

Cited by

References

    1. Albright TD, Desimone R. Local precision of visuotopic organization in the middle temporal area (MT) of the macaque. Exp Brain Res. 1987;65:582–592. - PubMed
    1. Allman J, Miezin F, Mc Guinness E. Stimulus specific responses from beyond the classical receptive field: neurophysiological mechanisms for local-global comparisons in visual neurons. Annu Rev Neurosci. 1985;8:407–430. - PubMed
    1. Angelucci A, Bressloff PC. Contribution of feedforward, lateral and feedback connections to the classical receptive field center and extra-classical receptive field surround of primate V1 neurons. Prog Brain Res. 2006;154:93–121. - PubMed
    1. Angelucci A, Bullier J. Reaching beyond the classical receptive field of V1 neurons: horizontal or feedback axons? J Physiol (Paris) 2003;97:141–154. - PubMed
    1. Angelucci A, Sainsbury K. Contribution of feedforward thalamic afferents and corticogeniculate feedback to the spatial summation area of macaque V1 and LGN. J Comp Neurol. 2006;498:330–351. - PubMed

Publication types

MeSH terms

LinkOut - more resources