In Part 1 of the experiment described herein, subjects wrote descriptions of the second panel of two-panel cartoons. Half of the cartoons were immediately preceded by the appropriate first panel and half were not. In Part 2, subjects described single-panel cartoons for one of two types of readers: 1, a person who could not see the cartoon or 2, a person looking at the cartoon with the subject but not 'getting it'. Both manipulations gave the predicted results. Subjects who saw panel 2 after panel 1 and subjects who could assume common perceptual information with the reader wrote shorter descriptions and used fewer definite articles than subjects who saw panel 2 only and subjects who could not assume common perceptual information. Results are used to argue against linguistic-object theories and for an interactive theory of the higher mental processes.