Wickens' release from proactive inhibition (PI) technique has been used to identify a number of encoding categories in short-term memory. When repeated presentation of words in one class followed by a shift to words in another class leads to improved recall (PI release), the classes are considered encoding categories. The Osgood semantic dimensions have been effective in producing PI release. These dimensions constitute a continuous semantic space, suggesting a corresponding encoding space with measures of PI release corresponding to distances. Using the Wickens technique, PI release scores were obtained in several regions of semantic space. Subjects were 40 undergraduates. The relation between distances in semantic space and release scores was highly linear, and the encoding space was approximately euclidean. Results suggest that the encoding process can use information from continuous variables in a quantitatively precise way.The release from proactive inhibition (PI) technique developed by Wickens (1970) has been used extensively to identify encoding categories in short-term memory (STM) (Wickens, 1970(Wickens, , 1972(Wickens, ,1973. Using the BrownPeterson paradigm, three trials with words in one class were followed by a fourth trial with words in a different class. Improved recall (PI release) on the shift trial implied that the classes were used as STM encoding categories, on the assumption that improvement in recall was due to differential encoding. (Wickens & Clark, 1968). The opposite ends of these dimensions were used as categories, although they really constitute continuous variables.Connotative meaning may be represented as a continuous semantic space, and it is used in STM encoding. Thus, it may be more appropriate in this instance to speak of an encoding space, rather than encoding categories. To be a useful conceptualization, encoding space should exhibit certain regularities. Release measures should closely correspond to shift distances in semantic This research was supported in part by a grant from the DAC, California State University, Fullerton. Portions of this research were presented at the meeting of the Western Psychological Association. Los Angeles, 1976. The author wishes to thank Richard H. Lindley for assistance in all phases of this research, and Thomas D. Wickens, who sponsors this paper and takes full editorial responsibility for its content, for helpful comments on the manuscript. Reprint requests should be sent to David G. Weeks, Department of Psychology, University of California, Los Angeles, 405 Hilgard, Los Angeles, California 90024. space. They should be at least monotonic, and ideally would be linearly related. The direction of shift should not make any difference. The position in the space (e.g., 1 to 2, 2 to 3) also should not affect release scores. The effect of a multidimensional shift should be predictable from the effects of the component unidimensional shifts. It should at least satisfy Equation 1:i where x is a distance derived from the release score, Xa is the deri...