and cones are probably active, as found for conventionally recorded oscillatory potentials (see [14] for review). Wu and Sutter [4] have shown that mfOPs are strongly dependent on the luminance of the stimulus. Reducing the luminance, and thus cone activity, caused a decrease in amplitude. On the other hand, a strong bleach, which eliminated rod activity, caused a reduction and delay in the implicit time of the first-order kernel responses and a complete elimination of all second-order (first-slice) potentials. Thus, a pronounced contribution of rod-cone interactions to the second-order kernel was proposed.

Examination of the topography of the second-order kernel (right in Fig. 1) reveals a superior-inferior asymmetry: Amplitudes in the superior retina are larger than those of the inferior. Rods are much more numerous in the superior retina than in the inferior [8, 9], so that this asymmetry provides further evidence for the importance of rods in the generation of mfOPs.

The second-order kernel (Fig. 1 right) also shows a pronounced nasotemporal asymmetry, with the nasal retina giving smaller responses than the temporal [15–17]. This asymmetry may be connected to the differing sizes of amacrine cell receptive fields in the nasal and temporal retinal hemifields [17]. However, Bearse and colleagues [16] have isolated an mfOP component that contributes to the nasotemporal asymmetry and is caused by activity arising from the optic nerve head, where ganglion cell axons first become myelinated. This signal is superimposed onto the activity stemming from the retina and is delayed on the temporal retina as a function of distance from the optic nerve head. Differences in amplitude in nasal and temporal fields will therefore be caused by the alignment or misalignment of the wavelets from the retina and those from the optic nerve head component.

Further evidence for the theory that mfOPs have two underlying components comes from pharmacological studies that have shown that the nasotemporal asymmetry disappears after intravitrial injection of tetrodotoxin citrate (TTX). This agent blocks sodium-dependent spiking at the ganglion cells, some amacrine cells, and the interplexiform cells [18].

THE INFLUENCE OF AGE AND GENDER

Age affects almost all visual functions, and mfOP recordings are no exception. In Fig. 2 we have reproduced representative traces from a healthy 20-year-old subject (dashed lines) along with those of a healthy 59-year-old (continuous lines). For analysis, the 61 traces recorded are grouped into five rings of hexagons concentric around the fovea. The first-order kernel is shown on the left and the first slice of the second-order kernel on the right. The first-order kernel potentials have the highest amplitudes between about 2° and 13° eccentricity as discussed in the preceding section (Fig. 1).

It will be seen that all potentials of both firstand second-order kernel are reduced in amplitude and delayed in the older subject compared to those of the younger. In a study involving 58 subjects and five decades of age, which controlled for prereceptoral factors such as pupil size and visual acuity, we found that the decrease in peak amplitude and the increase in implicit time are linear with age [19]. The rate of change with age was largest in the amplitudes of the second peak of the first-order kernel (0.4 nV/deg2 per decade) and the implicit time of the third peak of the second-order kernel (0.37nV/deg2 per decade). The amplitudes of the potentials have a larger distribution in the normal population than that of the implicit times, and combined with their generally small size, it can be difficult to obtain an adequate signal-to-noise ratio in older subjects.