The aim of the paper is to test the hypothesis that masking efficiency is related to the mask energy on the spatial frequency (SF) range critical to the task, irrespective on whether the mask is a noise
or a pattern. Two experiments were run inwhich targets werefragmented forms of objects masked atvarious intervals either by apattern orby thethreenoisemasks varying in the size of theirelements. In Experiment1,
in which participants had to discriminate the targetglobal shape (oval vs. round), masking efficiency was directly related to the mask energy on the low SFs critical to this task, whatever the mask. This
conclusion extends to the medium to high SF ranges critical to the naming task used in Experiment 2. These results are not compatible with the qualitative noise/pattern distinction used by Turvey (1973)
to operationalize his integration/interruption conception of masking. An interpretation in terms of early interactions between SF channels better account for the results.