Power spectrum model of visual masking: simulations and empirical data.

In the study of the spatial characteristics of the visual channels, the power spectrum model of visual masking is one of the most widely used. When the task is to detect a signal masked by visual noise, this classical model assumes that the signal and the noise are previously processed by a bank of linear channels and that the power of the signal at threshold is proportional to the power of the noise passing through the visual channel that mediates detection. The model also assumes that this visual channel will have the highest ratio of signal power to noise power at its output. According to this, there are masking conditions where the highest signal-to-noise ratio (SNR) occurs in a channel centered in a spatial frequency different from the spatial frequency of the signal (off-frequency looking). Under these conditions the channel mediating detection could vary with the type of noise used in the masking experiment and this could affect the estimation of the shape and the bandwidth of the visual channels. It is generally believed that notched noise, white noise and double bandpass noise prevent off-frequency looking, and high-pass, low-pass and bandpass noises can promote it independently of the channel's shape. In this study, by means of a procedure that finds the channel that maximizes the SNR at its output, we performed numerical simulations using the power spectrum model to study the characteristics of masking caused by six types of one-dimensional noise (white, high-pass, low-pass, bandpass, notched, and double bandpass) for two types of channel's shape (symmetric and asymmetric). Our simulations confirm that (1) high-pass, low-pass, and bandpass noises do not prevent the off-frequency looking, (2) white noise satisfactorily prevents the off-frequency looking independently of the shape and bandwidth of the visual channel, and interestingly we proved for the first time that (3) notched and double bandpass noises prevent off-frequency looking only when the noise cutoffs around the spatial frequency of the signal match the shape of the visual channel (symmetric or asymmetric) involved in the detection. In order to test the explanatory power of the model with empirical data, we performed six visual masking experiments. We show that this model, with only two free parameters, fits the empirical masking data with high precision. Finally, we provide equations of the power spectrum model for six masking noises used in the simulations and in the experiments.

Comparing the effect of the interaction between fine and coarse scales and surround suppression on motion discrimination.

Our ability to discriminate motion direction in a Gabor patch diminishes with increasing size and contrast, indicating surround suppression. Discrimination is also impaired by a static low-spatial-frequency patch added to the moving stimulus, suggesting an antagonism between sensors tuned to fine and coarse features. Using Bayesian staircases, we measured duration thresholds in motion-direction discrimination tasks using vertically oriented Gabor patches moving at 2°/s. In two experiments, we tested two contrasts (2.8% and 46%), five window sizes (from 0.7° to 5°), and two spatial frequencies (1 c/deg and 3 c/deg), either presented alone or added to a static pattern. When the moving pattern was presented alone, duration thresholds increased with size at high contrast and decreased with size at low contrast. At low contrast, when a static pattern of 3 c/deg was added to a moving pattern of 1 c/deg, duration thresholds were similar to the case when the moving pattern was presented alone; however, at high contrast, duration thresholds were facilitated, eliminating the effect of surround suppression. When a static pattern of 1 c/deg was added to a moving pattern of 3 c/deg, duration thresholds increased about 4 times for high contrast and 2 times for low contrast. These results show that the antagonism between sensors tuned to fine and coarse scales is more complex than surround suppression, suggesting that it reflects the operation of a different mechanism.

The L/M-opponent channel provides a distinct and time-dependent contribution towards visual recognition.

The visual pathway has been successfully modelled as containing separate channels consisting of one achromatically opponent mechanism and two chromatically opponent mechanisms. However, little is known about how time affects the processing of chromatic information. Here, parametrically defined objects were generated. Reduced-colour objects were interleaved with full-colour objects and measures of recognition performance (d') were compared by the continuous serial recognition paradigm. Measures were taken at multiple delay intervals (1, 4, 7, and 10 s). When chromatic variations were removed, recognition performance was impaired, but at the 1 s and 10 s intervals only. When luminance variations were removed, no impairment resulted. When only L/M-opponent modulations were removed, a deficit in performance was produced only at the 1 s and 10 s intervals, similar to the removal of chromatic variation. When only S-opponent modulations were removed, no impairment was observed. The results suggest that the L/M-opponent pathway provides a specialised contribution to visual recognition, but that its effect is modulated by time. A three-stage process model is proposed to explain the data.

Antagonism between fine and coarse motion sensors depends on stimulus size and contrast.

The perceived direction of motion of a brief visual stimulus that contains fine features reverses if static coarser features are added to it. Here we show that the reversal in perceived direction disappears if the stimulus is reduced in size from 2.8 deg to 0.35 deg radius. We show that for a stimulus with 1.4 deg radius, the reversals occur when the ratio between the contrast of the fine features and of the coarser features is higher than 0.8 and lower than 4. For stimulus with 0.35 deg radius, the reversals never appear for any contrast ratio. We also show that if the stimulus is presented within an annular window with small radius, errors disappear but they return if the radius is increased to 2 deg. The errors in motion discrimination described here can be explained by a model of motion sensing in which the signals from fine-scale and coarse-scale sensors are subtracted from one another (I. Serrano-Pedraza, P. Goddard, & A. M. Derrington, 2007). The model produces errors in direction when the signals in the fine and coarse sensors are approximately balanced. The errors disappear when stimulus size is reduced because the reduction in size differentially reduces the response of the low spatial frequency motion sensors.

Evidence for reciprocal antagonism between motion sensors tuned to coarse and fine features.

Early visual processing analyses fine and coarse image features separately. Here we show that motion signals derived from fine and coarse analyses are combined in rather a surprising way: Coarse and fine motion sensors representing the same direction of motion inhibit one another and an imbalance can reverse the motion perceived. Observers judged the direction of motion of patches of filtered two-dimensional noise, centered on 1 and 3 cycles/deg. When both sets of noise were present and only the 3 cycles/deg noise moved, judgments were reversed at short durations. When both sets of noise moved, judgments were correct but sensitivity was impaired. Reversals and impairments occurred both with isotropic noise and with orientation-filtered noise. The reversals and impairments could be simulated in a model of motion sensing by adding a stage in which the outputs of motion sensors tuned to 1 and 3 cycles/deg and the same direction of motion were subtracted from one another. The subtraction model predicted and we confirmed in experiments with orientation-filtered noise that if the 1 cycle/deg noise flickered and the 3 cycles/deg noise moved, the 1 cycle/deg noise appeared to move in the opposite direction to the 3 cycles/deg noise even at long durations.

Cone excitation ratios correlate with color discrimination performance in the horse (Equus caballus).

Six horses (Equus caballus) were trained to discriminate color from grays in a counterbalanced sequence in which lightness cues were irrelevant. Subsequently, the pretrained colors were presented in a different sequence. Two sets of novel colors paired with novel grays were also tested. Performance was just as good in these transfer tests. Once the horse had learned to select the chromatic from the achromatic stimulus, regardless of the specific color, they were immediately able to apply this rule to novel stimuli. In terms of the underlying visual mechanisms, the present study showed for the first time that the spectral sensitivity of horse cone photopigments, measured as cone excitation ratios, was correlated with color discrimination performance, measured as accuracy, repeated errors, and latency of approach.

Coherent motion perception fails at low contrast.

If the Fourier components of a moving plaid have similar temporal frequency, spatial frequency and contrast, coherent motion is perceived according to subjective judgements. We have devised a more objective method of determining the conditions required for coherent motion. Moving plaid stimuli were created with one stationary component. Plaids with a stationary component always have a single perceived direction of motion, which is determined by the presence or absence of coherent motion. In a temporal two-interval forced-choice paradigm we used a direction discrimination task to investigate the effect of varying the temporal and spatial characteristics of the Fourier components and pattern contrast on the probability of coherent motion perception. Agreement across observers regarding the conditions required for coherent motion was excellent using this more objective method. We find that patterns do not produce coherent motion when presented at contrast threshold, irrespective of how similar the Fourier components are. We also confirm that when the temporal frequency, spatial frequency and contrast of the gratings are sufficiently similar, observers report the direction of motion indicating coherent motion.

Spatial distribution of suppressive signals outside the classical receptive field in lateral geniculate nucleus.

A suppressive surround modulates the responsiveness of cells in the lateral geniculate nucleus (LGN), but we know nothing of its spatial structure or the way in which it combines signals arising from different locations. It is generally assumed that suppressive signals are either uniformly distributed or balanced in opposing regions outside the receptive field. Here, we examine the spatial distribution and summation of suppressive signals outside the receptive field in extracellular recordings from 46 LGN cells in anesthetized marmosets. The receptive field of each cell was stimulated with a drifting sinusoidal grating of the preferred size and spatial and temporal frequency; we probed different positions in the suppressive surround with either a large half-annular grating or a small circular grating patch of the preferred spatial and temporal frequency. In many of the cells with a strong suppressive surround (29/46), the spatial distribution of suppression showed clear deviation from circular symmetry. In the majority of these of cells, suppressive signals were spatially asymmetrical or balanced in opposing areas outside the receptive field. A suppressive area was larger than the classical receptive field itself and spatial summation within and between these areas was nonlinear. There was no bias for suppression to arise from foveal or nasal retina where cone density is higher and no other sign of a systematic spatial organization to the suppressive surround. We conclude that nonclassical suppressive signals in LGN deviate from circular symmetry and are nonlinearly combined.

Visual system: how is the retina wired up to the cortex?

A single retinal output neuron transmits to primary visual cortex through multiple pathways with different strengths. A new study in which activity was simultaneously recorded in pairs of retinal and cortical neurons provides evidence that these pathways converge on a single cortical neuron.

Visual mechanisms of motion analysis and motion perception.

Psychophysical experiments on feature tracking suggest that most of our sensitivity to chromatic motion and to second-order motion depends on feature tracking. There is no reason to suppose that the visual system contains motion sensors dedicated to the analysis of second-order motion. Current psychophysical and physiological data indicate that local motion sensors are selective for orientation and spatial frequency but they do not eliminate any of the three main models-the Reichardt detector, the motion-energy filter, and gradient-based sensors. Both psychophysical and physiological data suggest that both broadly oriented and narrowly oriented motion sensors are important in the early analysis of motion in two dimensions.

Gain control from beyond the classical receptive field in primate primary visual cortex.

Gain control is a salient feature of information processing throughout the visual system. Heeger (1991, 1992) described a mechanism that could underpin gain control in primary visual cortex (V1). According to this model, a neuron's response is normalized by dividing its output by the sum of a population of neurons, which are selective for orientations covering a broad range. Gain control in this scheme is manifested as a change in the semisaturation constant (contrast gain) of a V1 neuron. Here we examine how flanking and annular gratings of the same or orthogonal orientation to that preferred by a neuron presented beyond the receptive field modulate gain in V1 neurons in anesthetized marmosets (Callithrix jacchus). To characterize how gain was modulated by surround stimuli, the Michaelis-Menten equation was fitted to response versus contrast functions obtained under each stimulus condition. The modulation of gain by surround stimuli was modelled best as a divisive reduction in response gain. Response gain varied with the orientation of surround stimuli, but was reduced most when the orientation of a large annular grating beyond the classical receptive field matched the preferred orientation of neurons. The strength of surround suppression did not vary significantly with retinal eccentricity or laminar distribution. In the marmoset, as in macaques (Angelucci et al., 2002a,b), gain control over the sort of distances reported here (up to 10 deg) may be mediated by feedback from extrastriate areas.

The nature of V1 neural responses to 2D moving patterns depends on receptive-field structure in the marmoset monkey.

A plaid pattern is formed when two sinusoidal gratings of different orientations are added together. Previous work has shown that V1 neurons selectively encode the direction and orientation of the component gratings in a moving plaid but not the direction of the plaid itself (Movshon et al. 1985). We recorded the responses of 49 direction-selective neurons to moving gratings and plaid patterns in area V1 of the anesthetized marmoset monkey (Callithrix jacchus). The responses of V1 neurons to rectangular patches of varying lengths and widths containing gratings of optimal spatial frequency were used to measure size and aspect ratio of the receptive-field subunits. We measured responses to plaid patterns moving in different directions and graded the magnitude of the response to the direction of motion of the plaid and the response to the direction of motion of the component gratings. We found significant correlations between receptive-field structure and the type and strength of its response to moving plaid patterns. The strength of pattern and component responses was significantly correlated with the interrelated properties of direction tuning width (Spearman's r = 0.82, P < 0.001), and receptive-field subunit aspect ratio (Spearman's r = -0.79, P < 0.001). Neurons with broad direction tuning and short, wide receptive-field subunits gave their greatest response when the plaid moved in their preferred direction. Conversely, neurons with narrow direction tuning and long, narrow receptive-field subunits gave their greatest responses when the plaid moved in a direction such that one of its components moved in the preferred direction.

The uses of colour vision: behavioural and physiological distinctiveness of colour stimuli.

Colour and greyscale (black and white) pictures look different to us, but it is not clear whether the difference in appearance is a consequence of the way our visual system uses colour signals or a by-product of our experience. In principle, colour images are qualitatively different from greyscale images because they make it possible to use different processing strategies. Colour signals provide important cues for segmenting the image into areas that represent different objects and for linking together areas that represent the same object. If this property of colour signals is exploited in visual processing we would expect colour stimuli to look different, as a class, from greyscale stimuli. We would also expect that adding colour signals to greyscale signals should change the way that those signals are processed. We have investigated these questions in behavioural and in physiological experiments. We find that male marmosets (all of which are dichromats) rapidly learn to distinguish between colour and greyscale copies of the same images. The discrimination transfers to new image pairs, to new colours and to image pairs in which the colour and greyscale images are spatially different. We find that, in a proportion of neurons recorded in the marmoset visual cortex, colour-shifts in opposite directions produce similar enhancements of the response to a luminance stimulus. We conclude that colour is, both behaviourally and physiologically, a distinctive property of images.

Visual system: 'S' is not for saccades.

Coloured flashes that are visible only to the short-wavelength-sensitive S cones interfere with shifts of visual attention but not with shifts of gaze (saccades). Attention and gaze must therefore be directed by different visual sub-systems.

Feedback from V1 and inhibition from beyond the classical receptive field modulates the responses of neurons in the primate lateral geniculate nucleus.

It is well established that the responses of neurons in the lateral geniculate nucleus (LGN) can be modulated by feedback from visual cortex, but it is still unclear how cortico-geniculate afferents regulate the flow of visual information to the cortex in the primate. Here we report the effects, on the gain of LGN neurons, of differentially stimulating the extraclassical receptive field, with feedback from the striate cortex intact or inactivated in the marmoset monkey, Callithrix jacchus. A horizontally oriented grating of optimal size, spatial frequency, and temporal frequency was presented to the classical receptive field. The grating varied in contrast (range: 0-1) from trial to trial, and was presented alone, or surrounded by a grating of the same or orthogonal orientation, contained within either a larger annular field, or flanks oriented either horizontally or vertically. V1 was ablated to inactivate cortico-geniculate feedback. The maximum firing rate of LGN neurons was greater with V1 intact, but was reduced by visually stimulating beyond the classical receptive field. Large horizontal or vertical annular gratings were most effective in reducing the maximum firing rate of LGN neurons. Magnocellular neurons were most susceptible to this inhibition from beyond the classical receptive field. Extraclassical inhibition was less effective with V1 ablated. We conclude that inhibition from beyond the classical receptive field reduces the excitatory influence of V1 in the LGN. The net balance between cortico-geniculate excitation and inhibition from beyond the classical receptive field is one mechanism by which signals relayed from the retina to V1 are controlled.