Remembering Outside the Box

See allHide authors and affiliations

Science  03 Jul 2009:
Vol. 325, Issue 5936, pp. 40-41
DOI: 10.1126/science.1177156

The prevailing view in memory research is that the mammalian brain is composed of a number of heterogeneous modules, each responsible for a different cognitive function, including different types of memory. An emerging alternative view, however, suggests that instead of such modules, the brain is organized in terms of the multipurpose representations that different regions support. As such, a given representation—and thus a given brain region—could be useful for many different functions. Evidence for this latter view has focused largely on demonstrating that regions within the putative memory system of the mammalian brain also play a role in another high-level function: perception (14). On page 87 of this issue, López-Aranda et al. (5) show that regions within the brain's putative perceptual system also play a role in memory, further solidifying evidence for a representational view of brain organization.

The dominant textbook paradigm of human memory is the multiple memory systems view, according to which different kinds of memory, such as declarative memory (memory for facts and events) and nondeclarative memory (memory for skills and procedures), are subserved by different modules in the brain (6, 7). Declarative memory is the domain of the medial temporal lobe memory system, which includes structures such as the hippocampus and perirhinal cortex. This system can be contrasted with a separate perceptual representation system (8), which includes the extrastriate visual cortex (comprising roughly 30 regions including, V2, V3, V4, and IT in the primate brain). This perceptual system is, according to the standard view, critical both for representing the structure of visual stimuli and for nondeclarative memory functions such as priming (an increase in sensitivity to a stimulus due to prior experience).

However, it may be more useful to think about brain regions, not in terms of psychologically defined modules, but in terms of the representations that they contain (912) (see the figure). For example, in visual memory, structures within the medial temporal lobe (such as the perirhinal cortex) may be understood as an extension of the ventral visual object processing system—or the “what” pathway—which flows from the visual cortex to the medial temporal lobe (13, 14). This view suggests that structures in the “what” pathway and the medial temporal lobe should not be segregated according to their putative roles in perceptual versus mnemonic function. Instead, the entire stream is important for multiple functions including both memory and perception.

Brain organization.

The standard view in memory research suggests that distinct brain regions support different cognitive functions. The representational view of the human brain, by contrast, suggests that a given brain region could be useful for multiple high-level functions. For example, information represented in the “what” pathway flowing through the visual cortex to the medial temporal lobe may be important for both memory and perception.


A behavioral paradigm that has been critical in establishing the multiple memory systems view is object recognition memory. In this model, a subject is presented with an object to study. After a delay, the study object and a novel object are presented. A typical subject will preferentially select or spend more time investigating the novel object, and this is taken to be an index of memory for the study object. Object recognition memory is considered to be a canonical example of declarative memory and is therefore thought to depend exclusively on the medial temporal lobe memory system (15). However, López-Aranda et al. demonstrate the involvement of an area outside this system—specifically, cells that comprise layer 6 of the extrastriate visual cortical area V2. In their study, lesions of cells in layer 6 of the V2 region in the rat brain impaired object recognition memory, which suggests that V2 is critical to this function. This is not consistent with the prevailing multiple memory systems view, which would predict that damage to regions outside the medial temporal lobe should have little effect on object recognition memory. This finding is consistent, however, with the representational view, which would predict that damage to any region within the ventral visual cortex-to-hippocampal stream could affect object recognition memory.

It might be argued that damage to the V2 region may have cut off visual information to the medial temporal lobe. Such a disruption would render the medial temporal lobe memory system unable to function properly, thereby making the findings of López-Aranda et al. consistent with the modular view of memory. However, lesions made in the V2 region were highly selective in that an excitotoxin was used to damage only cell bodies in V2. Thus, neuronal extensions (fibers) would have been left intact, passing from V1 through V2, to downstream areas. Thus, visual information could still pass through V2. There were also plenty of “jumping projections” from neurons in information processing regions upstream of V2 to regions downstream from V2, including the medial temporal lobe.

More interestingly, López-Aranda et al. have provided further evidence to suggest that the involvement of V2 in object recognition memory cannot be interpreted simply in terms of a lack of information flowing from the visual cortex to the medial temporal lobe (due to V2 cell damage).When the protein RGS-14, which is involved in signaling pathways implicated in memory, was overexpressed in cells of layer 6 in V2, rats could remember more objects, and for much longer, relative to controls. The same manipulation within the medial temporal lobe (specifically, within the dentate gyrus and CA1 region of the hippocampus) had no effect on object recognition memory. Thus, López-Aranda et al. show that manipulations outside of the putative medial temporal lobe memory system can both impair and facilitate object recognition memory. This result is also consistent with previous studies reporting no effect of hippocampal manipulations on object recognition memory and is consistent with the view that the medial temporal lobe memory system is not a homogeneous system but contains distinct, dissociable components (16, 17).

An outstanding question is how molecules such as RGS-14 facilitate memory and what function they have in the mechanisms of cognition in the normally functioning brain. These are critical questions for future research.


Navigate This Article