# Abstract We show that visual interference impairs people’s ability to make use of visual knowledge. These results provide strong evidence that making use of stored visual knowledge—long-term memory of what things look like—depends on perceptual mechanisms. When tasked with answering questions about visual features of familiar objects, e.g., verifying that tables have flat surfaces, participants’ accuracy was reduced when viewing colorful noise patterns—a result most parsimoniously explained if the judgments required activation of visual representations that were being interfered with when viewing irrelevant patterns. Nonvisual knowledge was unaffected by the same visual interference manipulation. In a second set of studies, we show that the ability of words to cue visual knowledge is greatly reduced by presenting visual noise patterns during or after hearing verbal cues. That visual knowledge can be interfered with in this way demonstrates that it depends on mechanisms shared with visual perception. Although much of our conceptual content may abstract away from perceptual details, knowledge of what things look like appears to be represented in a visual format. # Experiments ## Property Verification Participants verified simple yes or no questions about common animals and objects. Some of the questions asked about the visual features of the objects (e.g., Does a swan have a long neck?). On half of the trials we presented visual interference during the questions. We hypothesized that the visual interference would impair performance on the visual feature questions but not equally difficult questions probing other kinds of (nonvisual) knowledge. ## Orientation Discrimination On each trial participants identified which of two otherwise identical images was correctly oriented. Before the images appeared, participants were cued with a word (e.g., "tree") or a burst of white noise. On half of the trials we presented visual interference prior to the target images. We hypothesized that the visual interference would interfer with the cueing effect---the benefit of a valid cue and the cost of an invalid cue.