Interruption of dot processing by a backward mask.

A white target field with 0-10 randomly positioned black dots was presented for 20 ms, and followed by a patterned mask with a duration of 200 ms. Subjects reported target numerosity, with strict or lax criteria. Numerosity functions, and control forced-choice discrimination results, supported an interruption model for backward masking by pattern, even for stimulus onset asynchronies as short as 50 ms. When the same targets were degraded by reducing their contrast, but not otherwise masked, results supported an integration model.