Linearity and Normalization in Simple Cells of the Macaque Primary Visual Cortex

By Matteo Carandini, David J. Heeger, J. Anthony Movshon. 1997

Simple cells in the primary visual cortex often appear to compute a weighted sum of the light intensity distribution of the visual stimuli that fall on their receptive fields. A linear model of these cells has the advantage of simplicity and captures a number of basic aspects of cell function. It, however, fails to account for important response nonlinearities, such as the decrease in response gain and latency observed at high contrasts and the effects of masking by stimuli that fail to elicit responses when presented alone. To account for these nonlinearities we have proposed a normalization model, which extends the linear model to include mutual shunting inhibition among a large number of cortical cells. Shunting inhibition is divisive, and its effect in the model is to normalize the linear responses by a measure of stimulus energy. To test this model we performed extracellular recordings of simple cells in the primary visual cortex of anesthetized macaques. We presented large stimulus sets consisting of (1) drifting gratings of various orientations and spatiotemporal frequencies; (2) plaids composed of two drifting gratings; and (3) gratings masked by full-screen spatiotemporal white noise. We derived expressions for the model predictions and fitted them to the physiological data. Our results support the normalization model, which accounts for both the linear and the nonlinear properties of the cells. An alternative model, in which the linear responses are subject to a compressive nonlinearity, did not perform nearly as well.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Citations in the corpus

Normalization as a canonical neural computation (2011, methodology): “Indeed, the normalization model was first developed to account for the physiological responses of neurons in the primary visual cortex (V1”
Generalisation in humans and deep neural networks (2018, background): “colour), image contrast and the amount of additive visual noise have been powerful means of exploring the human visual system, revealing much about the internal computations and mechanisms at work [42–48].”
A model of neuronal responses in visual area MT. (1998, positive): “Many of these behaviors can be accounted for by incorporating response normalization (Robson, 1988; Bonds, 1989; Albrecht & Geisler, 1991; Heeger, 1991, 1992a,b, 1993; DeAngelis et al., 1992; Carandini & Heeger, 1994; Carandini et al., 1997; Tolhurst & Heeger, 1997a,b; Nestares & Heeger, 1997).”
RULES OF COMPETITIVE STIMULUS SELECTION IN A CHOLINERGIC ISTHMIC NUCLEUS OF THE OWL MIDBRAIN (2011, unknown): “Normalizing inhibition serves to stabilize network activity and adjusts the sensitivity of neurons according to the average level of activity within a local pool of neurons, a computation that helps to avoid response saturation (Heeger, 1992; Carandini et al., 1997; Olsen et al., 2010).”
Synaptic Dynamics in Analog VLSI (2007, background): “Computational and biological studies have attributed different roles to shunting inhibition, such as logical AND-NOT) Koch, Poggio, & Torre, 1983), and normalization ( Carandini, Heeger, & Movshon, 1997 ) functions.”
The Normalization Model of Attention (2009, methodology): “, 2002; Priebe and Ferster, 2008) or feedback connections (Carandini et al., 1997; Heeger, 1993) or a combination of the two.”
Multisensory integration: flexible use of general operations (2014, unknown): “Recently, the divisive normalization model developed for visual processing (Carandini et al., 1997; Reynolds and Heeger, 2009), and described as a ‘‘canonical operation’’ (Carandini and Heeger, 2012), was shown to explain important features of multisensory integration such as inverse effectiveness a”
Comparing deep neural networks against humans: object recognition when the signal gets weaker (2017, background): “…1995), a mechanism called contrast gain control evolved, serving the human visual system as a contrast normalization technique by taking into account the average local contrast rather than the absolute, global contrast (e.g. Carandini et al., 1997; Heeger, 1992; Sinz & Bethge, 2009, 2013).”
Dynamics of Suppression in Macaque Primary Visual Cortex (2006, background): “The mechanism that was originally postulated to account for cross-orientation suppression involves a pooled “normalization” signal that arises from cells within striate cortex (Bonds, 1989; Heeger, 1992; Carandini et al., 1997).”
Coding of color and form in the geniculostriate visual pathway (invited review). (2005, background): “The contrast gain control model is a remarkably successful “fix” for the three main failures of the linear model, providing an accurate quantitative description of all three nonlinear effects.(182)”
Underlying principles of visual shape selectivity in posterior inferotemporal cortex (2004, result): “Thus, our more general models point toward a specific, biologically plausible mechanism analogous to successful models of neural summation in other domain”
Spatial Summation in the Receptive Fields of MT Neurons (1999, methodology): “No other work on MT has explicitly considered nonlinear summation, but in work on simple cells in V1 by Carandini et al. (1997), a related nonlinearity helps describe the summation of pairs of superimposed grating stimuli varying in contrast.”
Attentional modulation of visual processing. (2004, methodology): “This finding is illustrated in Figure 3C, which shows data recorded by Carandini et al. (1997) from a neuron in area V1 of the anesthetized macaque, when two spatially superimposed gratings, differing in orientation, appeared simultaneously within the receptive field (see also Bonds 1989, DeAngelis…”
Population coding under normalization (2009, methodology): “The normalization model we adopt is one that has been used widely to model the responses of simple cells in primary visual cortex (Carandini et al., 1997; Heeger, 1992):”
Neurophysiological investigation of the basis of the fMRI signal (2001, background): “In general, neurons show an increasing response that is nonlinear at low contrasts, and a linear increase up to a certain contrast level at which the response reaches its asymptot”
Neural mechanisms of orientation selectivity in the visual cortex. (2000, background): “This shunt would lower the membrane time constant of the cell, thereby lowering its integration time, with the resulting effect of advancing the phase of responses to sinusoidal gratings and enhancing responses to higher temporal frequencies (Carandini & Heeger 1994, Carandini et al 1997).”
A Canonical Neural Circuit for Cortical Nonlinear Operations (2008, unknown): “A similar mechanism was used to explain contrastdependent, sigmoid-like neural responses (Carandini & Heeger, 1994; Carandini, Heeger, & Movshon, 1997; Heeger, 1993) and center-surround effect (Cavanaugh, Bair, & Movshon, 2002; S´ eries, Lorenceau, & Fr´ egnac, 2003) in the primary visual cortex.”