As it turns out, addressing the issue of the continuity of confabulatory phenomena is less easy business as it might appear at first. A few decades ago, Berlyne lamented that the concept of confabulation had been “poorly defined and variously interpreted” (1972, 31), and even now the concern seems valid: the concept has been defined liberally and to refer to various collections of inaccurate statements, which have been classified according to the mode of elicitation (spontaneous vs. provoked), their content (fantastic vs. plausible), their stability and selectivity (stable vs. ephemeral), and even the domain in which they become manifest (memory, perception, action and emotion), thus leading one to wonder whether this should be best conceived of as a unitary construct or a somewhat arbitrarily selected hodgepodge of phenomena. Concerns over the possible heterogeneity of imperfect cognitions typically associated with clinical populations are not limited to the case of confabulation (Samuels 2009), but the latter is particularly interesting. If the concept of confabulation has been defined too liberally and used to refer to heterogeneous phenomena, it might be that phenomena displayed by clinical and non-clinical populations are actually not analogous. This paper systematically addresses the definitions of confabulation offered in clinical and non-clinical research and the cognitive domains in which phenomena reported as confabulatory fall in clinical and non-clinical contexts.