Disruption of neurotransmission in Drosophila mushroom body blocks retrieval but not acquisition of memory

476 NATURE | VOL 411 | 24 MAY 2001 | www.nature.com about the test grating (48 c.p.d.) on the frequency scale. The contrast of the 36 c.p.d. adapting grating was set to half of its threshold contrast, 3.5%. No measurable tilt after-effect was found under this condition (Fig. 2b). Similarly, in measurements of the elevation in contrast threshold produced by pre-exposure to a range of adapting grating contrasts, at three spatial frequencies spanning the resolution limit, we found that adapting gratings of frequency greater than the resolution limit were more effective than correspondingly subthreshold gratings of lower frequency (although subthreshold gratings in the high but resolvable frequency range were not entirely ineffective). Thus, the cortical requirement for conscious perception seems to depend on spatial frequency, and not only on contrast or signal strength represented at the cortical input. The idea that limits on visual resolution are partly imposed at the cortical level is supported by evidence that cortically projecting thalamic relay neurons in macaque often respond well to spatial frequencies far above the human resolution limit, in some cases exceeding 100 c.p.d. (ref. 15). If the projection from thalamus to cortex were as precise as the one from retina to thalamus, this would allow the visual system to form a representation of unresolvable patterns at the cortical site of pattern adaptation. The lower spatialfrequency limits for cortical after-effects (70 c.p.d.), as compared with thalamic neurons (100 c.p.d.), may re ̄ect imperfect precision in the projection from thalamus to cortex. In normal vision with incoherent light, diffraction markedly reduces the retinal image contrast for spatial frequencies near the resolution limit. Why should the cortex have orientation-selective mechanisms (or frequency-selective ones) that respond to high spatial frequencies that are normally only faintly represented in the retinal image? One answer derives from the view that neural mechanisms might compensate for optical blur: frequency components that have been optically attenuated might even require a reciprocal enhancement of neural sensitivity for their appropriate representation. But on our evidence, activation of orientation-selective units at the stage of cortical pattern adaptation is not suf®cient for perceptual awareness of the pattern orientation. The nature of the added requirement is not clear: one possibility is that information must be relayed from primary visual cortex to another region of the brain to be represented in conscious experience. M