The hemispheric encoding/retrieval asymmetry (HERA) model is a process-specific description of experimental data provided by a large set of functional neuroimaging studies. According to HERA, left prefrontal cortex (PFC) is more involved than right PFC in episodic memory encoding, whereas right PFC is more involved than left PFC in episodic memory retrieval. Recently it has been claimed that this description does not hold for nonverbal materials. Here we propose a more precise formulation of HERA than previously, and argue that there is sufficient evidence to conclude that HERA, as reformulated, is true for both verbal and non-verbal materials.It has been known for some time now that structures within the two cerebral hemispheres do not contribute equally to mental activity [1]. The best known form of such hemispheric asymmetry is language: the left hemisphere is more involved in verbal processing than is the right. A more recent addition to the list of hemispheric asymmetries, although limited to the frontal regions of the brain, emerged from early PET studies of memory, and had to do with component processes of memory. Data from numerous studies showed that during learning of new material (encoding), left PFC tended to be more active than right PFC, whereas during a subsequent recall or recognition test (retrieval), right PFC tended to be more active than left PFC. This general pattern of data, initially based only on a handful of studies [2,3], was dubbed the Hemispheric Encoding/Retrieval Asymmetry (HERA) model. Figure 1 illustrates the HERA asymmetry with data from a single study.Although exceptions to HERA have been reported, numerous PET and fMRI studies have confirmed the existence of the general HERA pattern (for reviews see [4 -7]). Recently, however, it has been suggested [8][9][10][11][12][13][14] that the asymmetry of PFC activations might reflect the nature of the materials involved in the comparison tasks (verbal versus non-verbal) rather than the memory processes (encoding versus retrieval). In light of these recent challenges, it is timely to re-evaluate the situation. We propose a more precise formulation of HERA, and argue that the challenges to it are based on confusion about what HERA is, and about what kinds of data are, and are not, relevant to it.
HERA revisedThe confusion is partly attributable to the looseness of the early formulation of HERA, and partly to subtleties of the relations involved. Here, in keeping with similar suggestions [10,15], we propose a stricter formulation of HERA. We suggest that the neuroanatomical data relevant to HERA should be based on a direct comparison of the two processes involved, with other variables, including materials, held constant.Let us abbreviate encoding as Enc, retrieval as Ret, a given site in left PFC as L, and the corresponding site in right PFC as R. And let us use the combination of tasks (Enc or Ret) and sites (L or R) to represent the level of (indirectly) measured neuronal activity at a given site in a given task.Then the encoding compon...