Remembering With a Stone-Age Brain

Ifmemory evolved, sculpted by the processes of natural selection, then its operating characteristics likely bear the ‘‘footprints’’ of ancestral selection pressures. Psychologists rarely consider this possibility and generally ignore functional questions in their attempt to understand how human memory works. We propose that memory evolved to enhance reproductive fitness and, accordingly, its systems are tuned to retain information that is fitnessrelevant. We present evidence consistent with this proposal, namely that processing information for its survival relevance leads to superior long-term retention—better, in fact, than most known memory-enhancement techniques. Even if one remains skeptical about evolutionary analyses, adopting a functional perspective can lead to the generation of new research ideas. KEYWORDS—evolution; memory; survival; fitness relevance Psychologists know a lot about human memory but very little about its function. It is well established that forming a visual image of an item improves its later retention, as does processing its meaning or generating the item initially from cues. Yet next to nothing is known about why these particular sensitivities developed or about the roles they play in actual functioning. Why did nature craft a memory system that is especially sensitive to imagery and the processing of meaning? Open any memory textbook and you will find little discussion about either the origins or functions of memory processes. Instead, researchers focus on structural (or proximate) mechanisms. When a memory effect is discovered—for instance, the discovery that forming a visual image benefits later retention of an item—it is ‘‘explained’’ by appealing to a set of general principles or processes—for example, elaboration (creating rich descriptions of the item in memory), contextual encoding (creating multiple representations of the item in memory), or the encoding–retrieval match (forming retrieval cues that are likely to be present in the retention environment). Thus, one might assume that visual imagery aids retention because it promotes elaborative encoding, produces dual mnemonic codes, or leads to especially accessible retrieval cues; self-generation of material from cues might encourage individual-item processing, making the traces of generated items easier to discriminate from nonoccurring items. Researchers seek to identify the general processes at work, much the same way a chemist analyzes a compound by breaking it down into constituent elements. The trouble with this kind of analysis, though, is that any complete understanding of structure—that is, the proximate mechanisms—is likely to demand some prior consideration of function. If memory evolved, shaped by the process of natural selection, then its structural properties should reflect their functionality (Tooby & Cosmides, 1992). Nature ‘‘selects’’ one physical design over another because that design has fitness value—it helps the organism solve an adaptive problem that, in turn, increases the chances of genetic transmission. The result is usually a tight fit between form and function; the selection pressure, or adaptive problem, constrains how and why the structure develops and the final form it takes. Thus, retinal cells are uniquely designed to process electromagnetic energy, the heart is uniquely designed to pump blood, and the kidneys are specially designed to help filter impurities. Analyzing these physical structures without reference to their function is inconceivable, yet a similar functional analysis is rarely applied to remembering (at least in the human domain—for some relevant animal work, see Domjan, 2005; Shettleworth, 1998). One could attempt to provide functional explanations for the mnemonic phenomena we mentioned earlier. For example, one might argue that cognition developed in the service of action, so embodied encodings (e.g., generating or forming a visual image) are beneficial to retention because they are congruent with the way our cognitive processes are designed to operate. However, post-hoc accounts—so-called ‘‘just-so stories’’—often represent the scourge of evolutionary analysis (Gould & Lewontin, 1979). Our laboratory has advocated a different approach, one that essentially starts from scratch. We have attempted to identify the selection pressures that may have shaped the Address correspondence to James S. Nairne, Department of Psychological Sciences, Purdue University, 703 Third Street, West Lafayette, IN 47907-2081; e-mail: [email protected]. CURRENT DIRECTIONS IN PSYCHOLOGICAL SCIENCE Volume 17—Number 4 239 Copyright r 2008 Association for Psychological Science evolution of memory, to generate a priori predictions, and then to design appropriate empirical tests. As we will illustrate, thinking in such a functional manner can lead to new and potentially rich avenues of empirical investigation. WHY DID MEMORY EVOLVE? The starting point of a functional analysis is to speculate about the adaptive problems that our memory systems evolved to solve. If we can identify those problems, or selection pressures, their ‘‘footprints’’ should be recognizable in the operating characteristics of the system. Identifying ancestral selection pressures— the so-called ‘‘environment of evolutionary adaptedness’’—can be a hazardous business, but we join others in believing that such an analysis is possible and heuristically useful (Hagen & Symons, 2007). At the very least, we can generate reasonable hypotheses about the likely (or unlikely) characteristics of any evolved memory mechanism. We offer three here. First, it is unlikely that memory and its associated mechanisms evolved simply to remember the past. There is little adaptive value in designing a system to recover the veridical past, given that the past can never occur again (at least in exactly the same form). Instead, our memory systems must be engineered to use the past in the service of the present, or perhaps to predict the likelihood of events occurring in the future (Suddendorf &Corballis, 1997; Tulving, 2002). The fact that memory is fundamentally constructive rather than reproductive, often laced with relevant but ‘‘false’’ recollections, provides prima facie evidence for this claim (Schacter & Addis, 2007). Second, evolvedmemorymechanisms are likely to be domainspecific, or sensitive to content; they should be tuned to remember certain kinds of information. A memory system that treats all environmental events the same would be maladaptive because not all events are equally important from a fitness perspective—for example, it is particularly important to remember the food source, the predator, or the appearance of a potential mate. Such tunings might develop with experience, but environments can be ‘‘clueless,’’ failing to deliver the necessary inputs, and selectivity is required in storage (Tooby&Cosmides, 1992). If we simply stored everything we encountered, ourminds would fill with clutter. We need a way to discriminate important from unimportant input, as well as mechanisms to clear away mnemonic clutter once it accumulates (M.C. Anderson, 2003). Third, and related to the second point, memory mechanisms should be geared especially to helping us perform actions that enhance our reproductive fitness. Again, memory did not develop in a vacuum; memory mechanisms evolved as design ‘‘solutions’’ to problems associated with fitness. Remembering the location of food, an activity preferred by a mate, or perhaps individuals who violate social contracts are likely to improve the chances of successful reproduction, which, in turn, sets the stage for structural modification via descent (Darwin, 1859). Table 1 provides a list of potential candidates for domain-specific mnemonic processes. We make no claims about the proximate mechanisms that might underlie such memory ‘‘tunings’’—such as separate systems or adaptations—but enhanced retention in situations such as those listed in Table 1 would likely confer a selection advantage. SURVIVAL PROCESSING ENHANCES RETENTION Despite more than a century of sustained laboratory investigations, researchers have little to say about how human memory operates in functionally relevant situations. J.R. Anderson and Schooler (1991) detected a strong correspondence between how information naturally recurs in the environment and standard forgetting functions, suggesting that retention depends on the likelihood that information will be reencountered and needed. Silverman and Eals (1992) suggested that women may be better equipped than men to remember information in fixed locales, perhaps because of how labor was divided during early environments of adaptation. Overall, however, functionally driven studies of human memory remain relatively rare. Our laboratory has begun examining several examples of ‘‘adaptive memory,’’ but we restrict our discussion here to survival processing. Given that reproductive fitness is contingent on survival, it is reasonable to hypothesize that our memory systems are specially engineered to retain information relevant to survival. As noted above, it should be easier for us to remember fitness-relevant information, things such as the location of food or the appearance of a predator, than fitness-irrelevant information. Empirically, there are several ways to test this ‘‘survival’’ hypothesis. For example, one could pick stimuli that seem inherently related to fitness and assess their mnemonic value. Is it easier to remember survival-relevant words such as corn,meat, or bear than it is to remember control words matched on other relevant dimensions? Is there an s-value (survival value) associated with a stimulus that predicts its memorability in the same way that imageability (concreteness) or word frequency predict retention? Perhaps, but survival relevance is likely to be context dependent. Food is survival relevant, but more so at the beginning of a meal than at its completion; a fur coat has high s-value at the North Pole, but low at the Equator. For this reason, we have focused primarily on how survival processing affects retention, much like one might ask how forming a visual image or assessing meaning affects retention. If an event is processed in terms of its survival value, from the perspective of a survival context, is that event subsequently remembered better? From an evolutionary perspective, domainspecific memory mechanisms are likely to have specific input criteria and to be activated only by specific cues. For instance, an evolved cheater-detection mechanism presumably operates only during the processing of social contracts, predator-avoidance systems in the presence of a predator, and mate-selection mechanisms in an appropriate social context. Certain events 240 Volume 17—Number 4 Adaptive Memory