by Pamela Effrein Sandstrom
Pamela Effrein Sandstrom is affiliated with the Helmke Library at Indiana University-Purdue University in Fort Wayne, Indiana. She can be reached by mail at the university at 2101 Coliseum Blvd. East, Fort Wayne, Indiana 46805; by e-mail at firstname.lastname@example.org or by phone at 219/481-6514.
Picture yourself a successful hunter-gatherer foraging in an uncertain environment, exploiting certain food resources and simply passing by others, all the while wary to avoid becoming prey yourself. In your daily forays you might sometimes forage like a generalist, consuming a wide array of resources. In other times and settings, you become more a specialist, ignoring plentiful foodstuffs in favor of some preferred prey type that, however rare, you invariably pursue and consume once you encounter it. What rules of thumb do human and animal subsistence foragers apply in choosing one resource and ignoring others? Why forage in one place and not another? How long should foragers remain in a particular microhabitat, and should they forage alone or in groups? What are the fitness consequences of different strategies?
Questions like these have absorbed anthropologists and biologists who take an evolutionary ecological approach to understand prey-predator-environment interactions. Their theoretical insights and empirical data offer a compelling model for librarians and information scientists studying information seeking and scholarly communication. We might ask: How and where do information seekers forage for valued resources? Are scholars, as a special category of information seekers, conforming to the same decision principles followed by all subsistence foragers? Are research specialties like productive microhabitats (or patches) within some larger information space? Do certain characteristics of resources and specialties explain or predict the use -- and deliberate non-use -- of information by scholars?
The many intriguing parallels between subsistence foragers and scholarly information seekers can be explored using an optimal foraging framework. (See resource list for recommended references on this topic.) In this deductive, microeconomic approach, the basic requirements of foraging models include an actor choosing among alternatives, a currency by which the actor measures costs and benefits, a set of constraints limiting behavior and a strategy set specifying the actor's range of available options. The currency (for example, calories in the food-foraging context) provides a measurement scale for assessing the effects of alternate decisions. By establishing some empirical cost-benefit currency, analysts are able to rank foragers' preferences, predict which resources will be pursued, and specify the net returns associated with particular choices.
But before approaching information seeking as a prey-choice or patch-choice problem, or proposing an appropriate currency for scholarly foraging, it is first helpful to know a good deal about a scholar's information environment.
Disciplinary Literatures as Bibliographic Microhabitats
The puzzle of how scholars' information behavior creates the skewness of bibliometric distributions -- concentrating some names, neglecting or scattering others -- intrigues scholars themselves as well as those of us charged with keeping systems of external memory. Literatures that are poorly connected to other bodies of work are often overlooked or ignored by information seekers, even when apparently relevant to their concerns.
Scholars spend years mastering the patterns of ideas and names that go together. Because of their training and active participation in a field, they are able to achieve a balance between familiar, core sources and unfamiliar, novel ones scattered widely throughout the published record. An individual's own knowledge or published body of work, however, rarely defines the breadth of the information base, even in a narrow specialty.
Instead, productive scholars maintain an awareness of the key authors in one or more central areas of research concern and demonstrate only nodding familiarity with work from more peripheral areas. Distinctly interdisciplinary work presents special information-foraging challenges. By operationalizing scholars' foraging behaviors through the empirical trace of their cited references, analysts are able to study the underlying intellectual structure and interrelatedness of research areas. These unobtrusive, remnant gestures of information seeking and use are on par with voting patterns, dimensions of which are often beyond the awareness or recollection of individuals. Through the graphical techniques of author co-citation analysis (ACA), specialties can be rendered as two- or three-dimensional maps of human interests and contiguous bibliographic microhabitats. ACA employs multidimensional scaling, cluster and principal components analysis in conjunction with behavioral and cognitive inventories, interviews and other validation measures. Authors of works judged important by a community of citing scholars and repeatedly juxtaposed in references-cited lists will cluster together. Authors whose works are rarely or never linked will stand apart. Clustered configurations of names often reveal shared schools of thought or methodological approach, common subjects of study, collaborative and student-mentor relationships, ties of nationality or other relationships -- all of which must be validated using independent methods.
ACA gives analysts the power to depict regions of scholarship as densely crowded and interactive, or the opposite, isolated and nearly vacant. ACA demonstrates its value as a tool to study information seeking when used within the framework of optimal foraging theory (OFT). Using insights from OFT, individuals' actual decisions to select or ignore information resources can be related to characteristics of particular resources and to features of the bibliographic microhabitats in which they forage.
Prey Choice and Other Optimal Foraging Models
The OFT prey-choice model predicts the composition of a forager's diet and the circumstances causing resources to be added or deleted. Analysts apply the model to natural environments containing a distribution of heterogeneous resources by separating foraging into mutually exclusive categories: search and handling activities. Searching involves the time a forager searches for all potential prey types before encountering the preferred prey. Handling begins with the forager's decision to pursue, and involves the time required to capture, process and consume one item of a given prey type.
In this deductive model, an individual's consumption mix represents the optimal set of prey types (that is, resources, in general terms) that personal and environmental constraints allow. For the scholar, these resources would include the range of authors or colleagues and their ideas and works consumed for the purpose of sustaining one's research and publishing career. OFT predicts that a particular resource will be included in the optimal diet if its energy return per unit handling time remains greater than the return rate (including search time) averaged for all higher-ranking resource types. This trade-off in time and energy is described by the intersection of inverse cost curves for searching and handling (as shown here).
Analysts plot time costs per unit harvest along the y-axis. Along the x-axis, resources are ranked according to their value in an appropriate cost-benefit currency per unit handling time. High-ranked resources are those with lower handling costs; higher handling costs are associated with low-ranking resources. Adding one more resource type increases handling costs but lowers searching costs, as more types are judged acceptable. An improvement in foragers' skills or technology also lowers searching costs.
The intersection of searching and handling cost curves describes the upper limit of resource types in the optimal diet. Specialism or generalism describes a forager's variability of feeding behavior and number of distinct resource types consumed. With many high-ranking resources available, foragers tend to have narrow diets, that is, they are specialists who pass by resources in favor of a few choice items. But when a favored type becomes depleted, the range broadens and foragers tend to be generalists. Foragers thus respond to changes in the density of high-ranking resources and constantly assess the average handling costs of items relative to the overall mix.
Similar to prey choice, the OFT patch-choice model is used to assess the time spent foraging in particular patches or microhabitats before moving elsewhere. This model applies when prey are superabundant (as is typical under conditions of information overload). In this situation, a forager's best strategy may be to sample patches but remain in the single microhabitat offering the highest return rate. Where resources become depleted through foraging, Charnov's marginal value theorem describes a forager's decision to abandon searching in one patch and exploit other locales. Again, a forager aims for a long-term average rate of return for all patches that maximizes energy but minimizes time spent foraging. Other OFT models relate resource distribution patterns to the costs and benefits of solitary foraging versus foraging in social groups.
Relating these principles to the scholarly forager demands that analysts regard a scholar's searching time as analytically distinct from handling time. Searching involves time spent identifying usable information. Scholars may invest time, for example, in searching via indexing and abstracting services or other bibliographic tools, or they may prefer routine reading, bibliographic chaining, monitoring key journals and semi-directed browsing. Searching also includes time consulting with colleagues about information resources. Scholars' information-seeking activities in the search phase fall along a continuum from active to passive, or from hunting-like to gathering-like strategies. Most scholars get by through a mix of monitoring and socially mediated strategies, avoiding the exhaustive search. In this sense, scholars appear to behave more as gatherers than hunters, preferring its reliable returns over more risky hunting strategies.
Information resources can be ranked like food items in terms of handling efficiency. Handling involves the time required to retrieve, read and process the contents of each encountered item. Information foragers incur lower handling costs by using a familiar set of resources (the same authors or specific items) over and over. This strategy has the effect of narrowing the overall variety of resource types in a scholar's consumption mix. But scholars also value recent publications or works from outside their major areas of research. Such resources should be admitted to the optimal diet if they can be handled efficiently relative to the overall mix of resources and if higher-ranked resources become less prevalent or devalued through overforaging.
Tracking Scholars' Foraging Behavior in the Wild
In reality, the relations among scholars, information resources and bibliographic microhabitats are dynamic. The shifts can be traced through ACA studies comparing different time periods, or captured at some single point in time and results compared for separate specialties. ACA provides a handy method for empirically tracking scholars in their information-seeking forays. OFT places the fascinating descriptive data of ACA studies in a wider explanatory framework.
One limitation of the OFT approach for studying scholars' behavior, however, is the problem of operationalizing a measurable currency. But the problem is not insurmountable. In analyzing how a group of anthropologists searched for and handled cited works by authors belonging to core and peripheral ACA clusters (Sandstrom, 1998, see references), I proposed a measurement scale that ranked the relative novelty of a resource to the research concerns of foragers. This novelty-redundancy continuum operates as the measurement scale against which information foragers gauged the costs (in foraging time) and benefits (in resource value) of their selection decisions.
At least two definitions of novelty are relevant to scholars in the OFT framework. The first relies on a conventional information-retrieval definition in the sense of personal unfamiliarity (an emic measure, that is, novelty is judged from the perspective of the subjects of a study). The study also measured the concept etically (that is, judged from the perspective of trained observers) by defining a resource's novelty in terms of its rarity or scarcity within a set of co-cited names. This scarcity-prevalence dichotomy is another way to describe the likelihood of encountering a particular pairing of authors in a particular bibliographic context.
For the study, I derived from a purposive sample of five active contributors the names of recently referenced authors and significant colleagues. I used these input names to create ACA maps of these scholars' research concerns. These renderings were based on information resources actually selected by individual contributors, but they also reflect the collective view of authors publishing in the Social Sciences Citation Index source journals. Cluster analysis classified co-cited authors into three center-periphery zones: the contributor's own cluster, other core clusters and omitted clusters. Results of this exploratory study showed that scholars' searching and handling mechanisms varied by zone, variations that were accounted for by the OFT models and traceable using ACA mapping.
I found that such searching behaviors as regular reading, browsing or the deliberate information foray (relatively solitary information-seeking activities) yielded resources belonging mostly to the peripheral zones of scholars' information environments. Peripheral resources tended to be first-time references and previously unfamiliar to citing authors. They were retrieved largely through temporary loan from colleagues or libraries -- both high-cost handling strategies. By contrast, resources belonging to core zones were comparatively low-cost in terms of handling, retrieved from existing personal collections or purchased. Core resources emerged from searching activities that emphasized routine monitoring of key sources and such socially mediated activities as graduate training, recommendation by a colleague or student, review of prepublication drafts and reprint exchange. These core resources had been referenced previously and the cited authors were often collaborators or acquaintances.
Findings suggest that visible and easily exploited resources belong to core ACA clusters. A scholar's regular reading and bibliographic chaining through cited references yields a background literature of familiar works and authors' names. A high mean co-citation rate within a cluster inhabited by a critical mass of productive contributors indicates that cluster members form an integral part of the information space. They are integral in the sense that they connect separate microhabitats of the mapped environment, knitting together parts of the field's intellectual structure.
Routine monitoring, informal communication and maintaining active social ties between producers and consumers facilitate information foraging in these high-density microhabitats. Both searching and handling costs are low because the information-supply system conveniently delivers a restricted range of resources. These resources are made even more accessible by forming part of the scholar's personal collections. The breadth of resources is thus narrower and foragers who concentrate on the items that these microhabitats deliver would be considered specialists.
In contrast, clusters with relatively low mean co-citation rates stand apart. Authors in these clusters are weakly connected to other cluster members and are not integral to the core research concerns of contributors to the specialty. Except for the chance juxtaposition of these authors with a few of the others on the map, this type of cluster exhibits a lower density of information resources of recognized value. The odds of encountering a given pair or set of authors are slight. These low-density microhabitats are characterized by a scarcity of familiar, previously paired authors. Low density signals that novel resources are more likely to be encountered than redundant ones.
Such microhabitats are difficult to exploit because the bibliographic and intellectual linkages are less salient. Successful foraging depends on labor-intensive reading in ever-wider circles, deliberate searching strategies and handling mechanisms that emphasize temporary access to materials over ownership. Maintaining social ties with individuals outside a scholar's core circles, however, facilitates information transfer from the periphery, making the handling of at least some peripheral resources less costly. Social interaction characterizes all sectors of the scholar's information environment but appears to differ in quality and quantity as distance between colleagues increases. Scholars who maintain many links to remote microhabitats have resource mixes of greater breadth and would be described more as generalists.
Further Research in Modeling the Bibliographic Topography
A research agenda within the socioecological ACA-OFT framework would aim to (1) characterize various information environments with greater precision, (2) establish more accurate decision criteria for information foragers and (3) predict and test how changed conditions alter the actual behavior of scholars. Diachronic studies (comparing shifts in co-citation structures over time) or synchronic comparisons between different fields or specialties hold significant promise for the empirical description of, and explanation for, scholars' elusive information-foraging behavior.
A Few Key References to Sources on Optimal Foraging Theory (OFT)
MacArthur, R. H., & Pianka, E. R. (1966). On optimal use of a patchy environment. American Naturalist, 100, 603-609.
Sandstrom, P. E. (1994). An optimal foraging approach to information seeking and use. Library Quarterly, 64, 414-449.
Sandstrom, P. E. (1998). Information foraging among anthropologists in the invisible college of human behavioral ecology: An author co-citation analysis. Ph.D. dissertation, Indiana University.
Smith, E. A. (1983). Anthropological applications of optimal foraging theory: A critical review. Current Anthropology, 24, 625-651.
Smith, E. A., & Winterhalder, B. (Eds.) (1992). Evolutionary ecology and human behavior. New York: Aldine de Gruyter.
Winterhalder, B. (1981). Optimal foraging strategies and hunter-gatherer research in anthropology. In B. Winterhalder & E. A. Smith (Eds.), Hunter-gatherer foraging strategies (pp. 13-35). Chicago: University of Chicago Press.