Ray Jackendoff's theory of conceptual meaning as outlined in his 1990 book (Semantic Structures), is an attempt to construct a semantic theory which is both psychologically realistic and compatible with the assumptions of generative linguistics. As such, it presents an implicit challenge to cognitive semantics, since it implies that linguistics can handle key phenomena (such as polysemy, metaphor, and the like) without abandoning generative assumptions. This review article presents a detailed summary and critique of Jackendoff's conceptual semantics. Key points include the following. (i) While Jackendoff claims to incorporate prototype effects within his theory, he does so only on a referential level. The theory is manifestly unable to account for family-resemblance phenomena within the lexicon. Detailed examination of the phenomena that Jackendoff discusses in 'Semantic Structures' reveals a variety of family-resemblance effects and patterns of polysemy, which 'Semantic Structures' either ignores or misanalyzes. (ii) While 'Semantic Structures' claims to incorporate the insights of Len Talmy's force dynamic account of causation, it actually offers a different, and arguably inferior account of causative patterns. (ii) While 'Semantic Structures' seeks to incorporate visual representations within the scope of lexical semantics, it can do so only in a very limited fashion because of its adherence to a strictly modular theory of cognition. Review of key semantic phenomena suggests that the role of visual information cannot be restricted in the fashion that Jackendoff postulates. (Paul Deane)