It is fundamental to vision science to measure people’s contrast thresholds for sinusoidal gratings or the responses of neurons in animals to gratings. Specialised graphics cards and displays costing thousands of pounds are needed for precise control of visual stimuli for research. Such costs are obviously prohibitive for teaching classes, but the psychophysics of grating thresholds allows excellent opportunities in the classroom to compare a person’s overall visual performance with the behaviour of the neurons supposedly underlying that behaviour.

We have written a program in Microsoft Visual Basic to run under Microsoft Windows on relatively modest ‘multimedia’ PCs, which are already in our class for Histology teaching. Although stimulus definition in space and time is not perfect, we have good approximation to more ‘professional’ displays. Furthermore, students can get fairly reliable estimates of their contrast thresholds quickly; some ‘professional’ experimental protocols require 100 s of stimulus trials, and 40 minutes’ concentration!

The observer faces the computer’s monitor and sees two square panes side-by-side. Each pane is 4 degrees square when viewed from 114 cm. At first, the panes are uniform mid-grey. The experimenter starts a trial (using the mouse) and a ‘movie’ begins either in the left pane or the right, chosen randomly by the computer. The movie is of a sinusoidal grating whose contrast gradually increases from zero over several seconds. The observer looks rapidly to and fro between the two panes and, when a faint stimulus is first seen in one pane, he/she pushes left arrow or right arrow appropriately on the keyboard. The movie stops and, if the correct pane was chosen, the computer displays the contrast at which the movie was interrupted; the experimenter can plot that value directly onto graph paper. There is no explicit penalty for anticipating and choosing the wrong pane, apart from the scorn of the observer’s peers! The protocol is quick and fairly reproducible; it appeals to the competitive instinct of some students.The gratings can be static or they can flicker at 5 Hz (180 deg phase-shift every 100 ms). Our displays have a frame rate of 60 Hz, and each ‘movie’ consists of 10 slightly different pictures per second. Under Microsoft Windows NT, we have not succeeded in synchronising the movie pictures exactly to the display frames. There are occasional glitches in movies of flickering gratings but these do not seem to detract.

Typically, computers display 256 grey levels, but we need many more than this to show the tiny nuances of luminance that compose a grating that is only just visible. Contrast thresholds can be less than 1 %. A crucial feature of our program is to extract over 3000 grey levels from the system; some of these are used to compensate for the nonlinear (roughly square-law) relation between the grey levels as calculated in the movie and the actual luminances on the display. The logical display resolution set by Windows must be the same as the physical pixel resolution of the display (1024 by 768 in our case) and the display set to ’32 bit’ colour. Each physical pixel can have 256 levels each of Red, Green and Blue (R, G and B), allowing 768 brightness levels (if we ignored colour). Starting from ‘black’, we could increment the brightness to ‘bright white’ by adding ones, in the sequence R-G-B-R-G-B-and so on; 2/3 of the brightness levels would not be precisely grey, and so it would be better to rearrange the RGB order every 3 grey levels. In fact, for moderate and low spatial frequencies, our movie frames are calculated with 128 by 128 logical pixels, but occupy 256 by 256 physical pixels on the display, each of which is less than 1 minute square. Then, each logical pixel occupies 4 physical ones, allowing a further factor of 4 in grey-level resolution. Increased grey levels are traded for decreased spatial resolution.

One can see how contrast thresholds for steady gratings depend upon spatial frequency: the pronounced low spatial-frequency cut supposedly shows lateral inhibition in the visual system or centre-surround antagonism of receptive fields. Flicker at 5 Hz shifts the sensitivity curve to low spatial frequencies, perhaps due to differences in the spatiotemporal preferences of P-and M-cells.

We can also examine ‘masking’. Trials begin with the two panes containing identical, visible gratings. When the ‘movie’ runs, the contrast of the grating in one pane will increase slowly or a second grating will be added to the mask. Students can see the Weber-Fechner relationship for contrast discrimination. They can also look at the orientation specificity of masking, which may reflect the orientation selectivity of neurons in striate cortex.