Functional magnetic resonance imaging (fMRI) was used to examine whether neural pathways used to encode pictures into memory were re-activated during retrieval of those memories. At encoding, subjects semantically classified common objects presented as pictures or words. At retrieval, subjects performed yes/no recognition memory judgments on words that had been encoded as pictures or as words. The retrieval test probed memory for the encoded item, but not memory for the modality of the encoded item (picture/word). Results revealed that a subset of the brain regions involved specifically in encoding of pictures were also engaged during recognition memory for the encoded pictures. Specifically, encoding of pictures relative to words engaged bilateral extrastriate visual cortex, namely fusiform, lingual, middle occipital, and inferior temporal gyri (Broadman area (BA) 18/19/37). Recognition memory judgments about words that were encoded as pictures relative to those that were encoded as words activated fusiform and inferior temporal gyri primarily in the left hemisphere. Thus, cortical areas originally involved in perception of a visual experience become part of the long-term memory trace for that experience. These findings suggest a neural basis for encoding specificity and transfer appropriate processing in human memory.
What is remembered depends upon how it was recorded into memory. This idea is central to two principles of memory that have broad behavioral support, encoding specificity and transfer appropriate processing. Encoding specificity states that encoding operations determine storage which in turn determines the effectiveness of retrieval-cues [29]. Transfer appropriate processing states that memory is enhanced to the extent that encoding operations are recapitulated at retrieval [19]. Thus, both principles predict an overlap between encoding and retrieval processes, psychologically, and by extension, in the brain. Specifically, both theories predict that brain regions activated during encoding ought to be re-activated during episodic retrieval.
Functional neuroimaging has visualized the overlap between brain regions underlying encoding and retrieval of the sensory modality of those encoding operations [22] and [33]. Subjects studied visual–auditory cue–target pairs and recalled the auditory information in response to the visual–cue at test. Regions in auditory cortex activated during encoding of auditory target stimuli were re-activated during visually cued recall. Functional imaging of cross-modality encoding and retrieval, therefore, provides evidence for the test-phase recapitualtion of modality-specific encoding operations.
Evidence for the neural overlap of encoding and retrieval operations within the same sensory modality remains inconclusive [18], [20], [21], [23] and [31]. In these studies, study and test stimuli comprised the same materials (e.g. study-faces, test-faces) or evoked the same stimulus attributes (study-object location, test-objects in two locations). Thus, encoding operations (e.g. perception of faces and location) performed at study were performed again during the retrieval test. Therefore, encoding-related regions that were activated during the study-phase would also be activated during the test-phase. Indeed, activation patterns for non-studied words in a recognition memory test were the same as those observed for encoding words in the study-phase [2]. Thus, in within-modality encoding and retrieval tests, it is impossible to separate activation associated with encoding of test-phase stimuli from that associated with the recapitulation of encoding operations.
One study that allowed for separation between activation due to encoding of test-phase stimuli and recapitulation of encoding operations failed to find conclusive evidence for overlap in brain regions underlying encoding and retrieval [13]. Subjects studied words and pictures and performed a recognition memory task using four types of retrieval-cues, study-phase words, pictures corresponding to study-phase words, study-phase pictures, and words corresponding to study-phase pictures. Visual materials presented at study differed from those at test in two retrieval conditions, pictures corresponding to study-phase words and words corresponding to study-phase pictures. These retrieval conditions, therefore, allow examination of the extent to which brain regions involved in material-specific encoding operations were re-activated during retrieval. Results revealed that brain regions associated with material-specific encoding were not activated significantly during retrieval. Studies of within-modality encoding and retrieval, therefore, have not visualized successfully the overlap in brain activation between encoding and retrieval.
We used functional magnetic resonance imaging (fMRI) to examine the extent to which picture-specific encoding regions were re-activated during episodic retrieval. At study, subjects encoded objects presented as pictures or words. At test, recognition memory was tested only with words that had been encoded as pictures or as words, or were novel. Thus, all test-phase stimuli were words and differed only in how they had been encoded. Therefore, observed activation differences between words that had been encoded as pictures and those that had been encoded as words could only be attributed to their material-specific mnemonic history rather than ongoing perception at test.
