Human Visual Mechanism Research Articles

A purely binocular mechanism in human vision

After scanning a bipartite field, the top half containing a grating tilted to the left: the bottom, a grating tilted to the right; an objectively co-linear pair of gratings will appear bent in the opposite direction: top to the right, bottom to the left (the tilt after-effect: TAE). Following monocular exposure, the binocular TAE is significantly smaller than the monocular TAE. If a weighted average of all activated channels produces the binocular TAE, then the reduced magnitude of that TAE suggests that binocular testing activates a process not adapted during monocular exposure.

A transducer function for threshold and suprathreshold human vision.

A nonlinear function is derived to describe the contrast transduction process for human visual mechanisms. This function is sigmoid in form, having an accelerating nonlinearity at low contrasts and a compressive nonlinearity at high contrasts. The resulting formulation is consistent with both signal detection theory and with Quick's (1974) equation for probability summation. Similarities between the present description of human vision and properties of complex cells in cat visual cortex are noted.

Spatial characteristics of movement detection mechanisms in human vision. II. Chromatic stimuli.

Visual detection of targets moving against structured background fields has been studied with near-monochromatic stimuli, selected so as to isolate the different increment threshold spectral response mechanisms. It is shown that for foveal vision, the red- and green-sensitive mechanisms (pi 5 and pi 4 respectively) yield IMG functions (Barbur and Ruddock, 1980), similar to those found with white light. In contrast, the blue-sensitive (pi 1) mechanism yields a low-frequency IMG response quite unlike that found for the other mechanisms. There is also considerable variation between subjects in this case. Measurements taken 30 degrees off-axis with low (1.4 log trolands) background illumination level, yield a low frequency response IMG function for both rod and cone spectral mechanisms, similar to those found with white light stimuli. At high illumination levels (> 2.2 log trolands) the IMG function for the pi 5-mechanism is shifted to higher spatial frequencies, as is also observed with white light stimuli. A wavelength-selective binocular interaction effect, manifested in the detection of moving targets, is also described, and it is suggested that this may be of value in the study of defective colour vision.

Spatial characteristics of movement detection mechanisms in human vision. I. Achromatic vision.

Threshold illumination levels, It, for visual detection of white light targets, moving across spatially structured background fields, have been measured and it is shown that, for a given background illumination level, It depends upon the spatial characteristics of the background field structure. Thus, with the background composed of a square-waveform grating of fundamental spatial frequency f cycles/deg, It is maximum for an intermediate value of f, and falls as f increases or decreases from this value. The relationship between It and f characterizes the interaction between movement detection and the background grating, and is designated the IMG function. The parametric properties of the IMG functions are described and it is established that the mechanisms which give rise to these functions are sensitive to the movement, but not the spatial structure of the target. They correspond, therefore, to the movement-sensitive Y-type mechanisms, observed in electrophysiological studies of cat and primate visual pathways. The spatial distribution of sensitivity associated with the IMG functions has been computed by 2-D transform methods, the computation yielding circularly symmetric, centre-surround antagonistic "receptive field" distributions.

Spatial characteristics of movement detection mechanisms in human vision. III. Subjects with abnormal visual pathways.

Spatial characteristics of movement detection mechanisms in human vision. III. Subjects with abnormal visual pathways.

Multiple bandpass filters in image processing

A system of bandpass filters, modelled on the early mechanisms of human vision, is described which provides a simple means to eliminate blur and noise from an image. The system automatically adjusts itself to suit local signal conditions in an image, without prior knowledge of signal statistics.

Sustained and transient mechanisms in human vision: Temporal and spatial properties

In two experiments, properties of sustained and transient mechanisms were studied psychophysically. In the first contrast thresholds were measured for 6 sinewave gratings ranging from 0.375 to 12.0c/deg at 10 durations ranging from 18 to 3000msec. Thresholds were measured in the presence and absence of high contrast 20msec gratings which masked the onsets and offsets of the signals. At 1.5 c/deg and above, the unmasked thresholds decreased as power functions of duration in two stages, reaching an asymptotic level near 1000msec. Below l.5c/deg, the unmasked threshold became independent of duration beyond 100 msec. At all frequencies, the masked thresholds decreased as power functions of duration to 1000msec or more, but the curves for 0.375 and 0.75c/deg never reached the unmasked asymptotic level. In the second experiment, spatial frequency bandwidths were obtained for sinewave gratings ranging from 0.375 to 12.0c/deg. by measuring threshold elevation as a function of the spatial frequency of masking gratings. At 3.0, 6.0 and 12.0 c/deg, the bandwidth functions peaked at the signal frequencies and showed medium bandwidth frequency selectivity. Below 1.5 c/deg, the bandwidth functions exhibited broader spatial frequency tuning, were of higher magnitude, and there was a shift in peak masking to frequencies near 1.0–1.5 c/deg which were above the signal frequencies. The results of both experiments are discussed in terms of the sustained/transient dichotomy.

Properties of length-selective and non-length-selective adaptation mechanisms in human vision.

The contrast threshold elevation effect has been measured for one dimensional (grating) and for two dimensional (spot) stimulus patterns. It has been shown previously (Burton and Ruddock, 1978) that such stimuli elicit, respectively, non-length-selective and length-selective adaptation effects. It is established that, unlike the frequency shift effect, the contrast threshold elevation effect is sensitive to the width of the light but not to that of the dark elements of the stimulus patterns. Adaptation to spot patterns elicits a significant threshold elevation for detection of both spot and grating test stimuli, but only under monoptic viewing conditions. The experimental findings are summarized in a block-diagram and it is shown that adaptation to grating patterns is successfully described by the spatial frequency response data given by Maudarbocus and Ruddock (1973).

The nature of the pi1 colour mechanism of W.S. Stiles.

The ppi1 colour mechanism was isolated by means of Stile's auxiliary field technique: that it is pi1 which is isolated is proven by both test and field action spectra. 2. The fundamental mechanisms of human trichromatic colour vision must satisfy Grassman's law of additivity. The hypothesis that pi1 is one of the three fundamentals is tested by experiments in which pairs (u1, u2) of monochromatic adapting fields are mixed 3. When two fields with wave-lengths in the neighbourhood of its primary mode (mu1, mu2 less than or equal to 500 nm) are combined, pi1 is field-additive, consistent with the hypothesis that this portion of the pi1 action spectrum is that of the short-wave-length-sensitive photoreceptors. 4. When a short-wave-length adaptation field (mu1 less than or equal to 500 nm) is mixed with a longer wave-length field (mu2 greater than or equal 550 nm) i1 is strongly non-additive. This result proves that the long-wave-length portion of the pi1 field spectrum is generated at least in part by a signal originating in the long- or middle-wave-length sensitive cones. 5. Analysis of the additivity failures supports a model of pi1 in which the signal to be detected is generated in the short-wave-length cones, and must pass serially through two gain stages: the gain in the first stage is controlled by the short-wave-length cones alone; the gain in the second stage is controlled by a signal originating in the middle, or long-wave-length cones, or both.

Sensitivity of control parameters in a model of saccadic eye tracking and estimation of resultant nervous activity

A model for the extraocular plant of the human visual eye tracking mechanisms is discussed. Its sensitivity to variation of controller signal nervous activity is studied in order to determine the type of activity that yields realistic simulations characteristic of typical saccadic eye movements.

Contrast sensitivity of individual colour mechanisms of human vision.

1. Contrast sensitivity functions of isolated colour mechanisms were measured at spatial frequencies from 0-2 to 32 c/deg. The contrast sensitivity vs. spatial-frequency functions of the red (pi5) and green (pi4) mechanisms are similar, while the blue (pi3) mechanism has lower absolute sensitivity and lower resolving power. Isolation of a single mechanism never increases its maximum sensitivity. 2. The shape of the contrast sensitivity function of a colour mechanism is established within the mechanism. Little if any inhibitory interaction takes place among colour mechanisms. 3. Differences that have been reported between the sensitivities of the red and green mechanisms, as well as the apparent "supersensitivity" of the isolated green mechanism, may be artifacts that result from the extrapolation procedures that were used to estimate the absolute sensitivities of the colour mechanisms.

Inhibition and disinhibition of direction-specific mechanisms in human vision.

MOTION-SENSITIVE mechanisms in human vision are selective for direction of stimulus movement. Psychophysical experiments reveal direction-specific channels which can be selectively desensitised or adapted1,2. At threshold these direction-specific mechanisms operate independently; they have little or no sensitivity for the opposite direction of motion3–6. Independence at threshold does not, of course, preclude interaction at suprathreshold stimulus levels. Indeed, inhibition has been found above threshold between otherwise independent spatial frequency-specific mechanisms7,8, orientation-selective mechanisms9,10, and binocular disparity-specific mechanisms11. We report here comparable measurements of inhibition between direction-specific channels. We also show that in appropriate conditions the inhibition can itself be reduced or eliminated, a disinhibition effect.

Adaptive filter mechanisms in human vision.

As a result of our experiments we find that the absolute visibility of simple edge and bar patterns is determined by a cooperation between the maximum amplitude and the difference between the maximum gradients occurring in the patterns. Proceeding from the assumption that the second spatial derivative represents a basic mechanism in our visual system (Mach-bands, effects of masking between narrow lines and sharp edges, existence of even-symmetrical receptive fields in the retina and the LGN), we suggest a nonlinear summation device which produces the difference between gradients as a sum over the second derivative. Expressed in a slightly different manner we assume that the sum over the excitation embodied in a Mach-band acts as criterion during the absolute visibility of edge and bar patterns.

Contrast response of human visual mechanisms sensitive to orientation and direction of motion

A slowly moving grating of low spatial frequency differentially raises the threshold of various kinds of test targets. Systematic variation of the contrast of the adaptation grating separates two distinct components of this response. One component depends upon the grating's orientation and is a power function of contrast. This component is independent of the grating's movement. The other component is controlled by the direction of movement of the grating and is sensitive only to a restricted range of low contrasts.

Size-Detecting Mechanisms in Human Vision

Inspecting a pattern of alternating dark and light bars makes it difficult to see a similar pattern presented afterward. This phenomenon can be used to isolate mechanisms responsive to bars of a given width. Our results suggest that the human visual system contains several different classes of size detectors, each maximally sensitive to visual targets with sizes in a particular range.

Size-detecting mechanisms in human vision.

Inspecting a pattern of alternating dark and light bars makes it difficult to see a similar pattern presented afterward. This phenomenon can be used to isolate mechanisms responsive to bars of a given width. Our results suggest that the human visual system contains several different classes of size detectors, each maximally sensitive to visual targets with sizes in a particular range.

Selectivities of human visual mechanisms for direction of movement and contour orientation.

The luminance threshold for horizontal contours which move upward is raised by prior exposure of any of wide range of different directions of movement. The threshold for a stationary grating of horizontal bars is raised only by prior exposure of similarly oriented bars. It is also found that sensitivity to changes in orientation of a stationary grating is greater than sensitivity to changes in the direction in which a grating moves. It is proposed that the interactions between pre-exposed adapting stimuli and the stimuli for which thresholds are determined reflect the orientation and direction selectivities of the underlying mechanisms.

Human color vision and color blindness.

The work of the past few years has begun to give us a reasonably complete account of receptor mechanisms in human vision. The human retina, like that of most other vertebrates, contains rods and cones. Human rods everywhere in the retina seem quite typical; but human cones present peculiar features. It has long been recognized that the outer segments of foveal cones are rod-shaped; they are attenuated cylinders about 50 microns long and about 1–2 microns wide.

ELECTRO-OPTICAL STIMULATION OF THE HUMAN RETINA AS A RESEARCH TECHNIQUE.

The subject of measuring transfer characteristics of the foremost mechanisms of human vision by simultaneous electrical and optical stimulation of the eye is introduced by a discussion of spatial interactions, the Brucke-phenomenon, non-linear highpass behaviour and linear lowpass behaviour.