R. Le Grand, C. J. Mondloch, D. Maurer, H. P. Brent

Nature 410, 890 (2001)

There is a mistake in Table 1 and its legend, as the result of a coding error. The final two sentences of the legend should read: “For adults, inversion reduced the accuracy of the configural set significantly more than that of the featural set (significant interaction between orientation and stimulus set (F 1,25 = 13.13, P < 0.01), followed by analyses of simple effects (F 1,27 = 62.57, P < 0.01)). Patients were less accurate than controls (matched for age, gender, race and handedness) only for the upright configural set (significant interaction between group, orientation and stimulus set (F 1,26 = 21.92, P < 0.01)).” Values in the 'featural, upright' column should be 80, 89 and 85 for adults, controls and patients, respectively. This error does not affect our conclusions.