header
Vol. 12 No. 4, October 2007

 

Invited paper

What is browsing— really?
A model drawing from behavioural science research


Marcia J. Bates
Department of Information Studies, University of California, Los Angeles, Los Angeles, California 90095-1520, USA


Abstract
Introduction. It is argued that the actual elements of typical browsing episodes have not been well captured by common approaches to the concept to date.
Method. Empirical research results reported by previous researchers are presented and closely analysed.
Analysis. Based on the issues raised by the above research review, the components of browsing are closely analysed and developed. Browsing is seen to consist of a series of four steps, iterated indefinitely until the end of a browsing episode: 1) glimpsing a field of vision, 2) selecting or sampling a physical or informational object within the field of vision, 3) examining the object, 4) acquiring the object (conceptually and/or physically) or abandoning it. Not all of these elements need be present in every browsing episode, though multiple glimpses are seen to be the minimum to constitute the act.
Results. This concept of browsing is then shown to have persuasive support in the psychological and anthropological literature, where research on visual search, curiosity and exploratory behaviour all find harmony with this perspective.
Conclusions. It is argued that this conception of browsing is closer to real human behaviour than other approaches. Implications for better information system design are developed.


Introduction

Though often seen as a casual, incidental behaviour in the general society, browsing, in the information world, is widely recognized as an important information seeking technique. In an academic context, scholars have argued that frequent browsing is often the only way to locate information and resources that cannot be readily described by index terms. Further, some kinds of information are recognized as relevant only upon discovery. In short, there are the things you know you do not know and the things you do not know you do not know. Browsing provides an alternative strategy for locating information of the first kind and may provide one of the crucial ways for information of the second kind to be encountered.

Information seeking is, after all, about finding out things that one does not know before the search begins. Therefore, uncertainty always accompanies the process to a greater or lesser degree. Sometimes it is possible to specify the information need very closely— "What was the population of Turkey in 1960?"— but other times specification in advance is impossible, as when the ugly duckling was surprised to discover serendipitously that he was actually a swan.

Much has been written about browsing in the library and information science literature, but it has generally been found difficult to specify browsing too closely, because 1) the conditions under which browsing is used vary widely, 2) it seems to be rather unpredictable in its very nature and 3) it seems to be employed in both more and less directed, intentional ways. In Accessing and Browsing Information and Communication, by Rice et al. (2001) the source in which browsing appears to have been most comprehensively reviewed and discussed, the authors find definitions of the following sort to be common: "Herner (1970) derived three… categories: 1) directed browsing, 2) semidirected or predictive browsing and 3) undirected browsing" (p.177).

After reviewing several of these sorts of definitions, the authors conclude that four dimensions have been most prominent in discussion of the subject (Rice et al. 2001):

  1. "The act of scanning. Browsing embodies an act of scanning, which has been variously described as looking, examining, or sampling where the person's body or eyes move smoothly at will" (p. 178).
  2. "Presence or absence of a purpose.… Although a purposeful act can be goal-directed or non-goal-directed, the presence of an intention suggests that the concept of browsing cannot be adequately described by behavioural characteristics alone" (p. 178).
  3. Search criteria may be extensively, minimally, or not at all specified. "At one end of this continuum, the objective of browsing is well defined; at the other end, the objective of browsing is not defined" (p. 178).
  4. The browser's knowledge of the resource browsed— either search paths in or contents of— can affect the quality of the interaction with the resource substantially (pp. 178-9).

The above conclusions were based on a review of the literature of library user studies. The authors then continued by reviewing six other fields that have some reason to be interested in this general sort of behaviour: 1) End-user computing and information science; 2) consumer research, 3) audience research, 4) organizational research— environmental scanning, 5) organizational research— informal communication and 6) environmental planning and architectural design (Rice et al. 2001: 182-210). The key insight for our purposes from this extensive review is that the authors found that the core cognitive and behavioural activity involved in browsing in these several literatures was largely the same as found in library user studies. In five of the six fields, the authors identify "scanning" as a part of the definitions of browsing applicable in that field. Scanning was missing only in their review of the literature of "organizational research— informal communication." There, Rice et al. define browsing thus: "… the notion of browsing in this context refers to casual access to social links and unpredictable exposure to many possible social interactions and to the implied knowledge available through these interactions" (Rice et al. 2001: 203).

We generally know intuitively what browsing is, because we have engaged in it ourselves and observed it in others. However, if we did not have that intuitive knowledge, we might have to conclude, based on the four points above (Rice et al. 2001), that browsing, at a minimum, consists of intentional visual scanning, with or without a well-defined objective and with or without fore-knowledge of the resource(s) browsed.

In this article, I would like to consider browsing more deeply. First, does browsing, at its core, only consist of intentional visual scanning, "where the person's body or eyes move smoothly at will" (Rice et al. 2001: 178)? Is there more to it?

Second, why do we browse at all? After all, there are many kinds of searching, known to trained information personnel, that the average untrained person does not engage in— but browsing seems to be widespread. Where does this impulse come from? How do people even know to do it?

Browsing in information studies

Our principal objective here is to understand browsing in the context of information behaviour and for the purposes of information science. However, it is valuable for us to know, based on the Rice, et al. review (2001) that something like browsing appears in the research of other disciplines, such as consumer research and organizational theory. Such a result suggests that browsing is a more general behaviour, something human beings do under a variety of impulses, not only in moments of information need. We will come back to this more general perspective in a later section.

Here, however, I wish first to challenge the common perspective, shared by Rice, et al. (2001), that browsing is a matter of visual scanning, "where the person's body or eyes move smoothly at will" (p. 178). If we imagine literally doing what this clause says, then we visualize the proverbial fur-trapper or scout in the woods or on the edge of a meadow, shading his eyes with a flattened hand held perpendicularly out from the forehead and rotating the head from one side to the other in a smooth scan of the environment.

The term "to scan" itself requires elaboration. The first three and most relevant, definitions of the verb in the American Heritage Dictionary are these:

1. To examine closely. 2. To look over quickly and systematically: scanning the horizon for signs of land. 3. To look over or leaf through hastily: scanned the newspaper while eating breakfast ("Scan." American Heritage Dictionary 2000).

The dictionary's usage note explains that the verb to scan has historically had two "opposite senses"— one "to examine closely" and the other "to look over quickly." It explains that the latter meaning is more recent and is now fully established. Rice et al. (2001) take advantage of this dual— and, arguably contradictory— set of meanings to claim that browsing includes both quick looks, "glancing around" (p. 218), as well as lengthier examination of resources: "scanning within a potential resource for assessment tends to be more thorough… " (p. 219). The authors base their understanding of the cognitive and behavioural process of browsing on this central idea of scanning.

They elaborate as follows:

… [T]he notion of browsing has been construed as a shopping activity, a viewing pattern and a screening technique in addition to a search strategy. In the most fundamental sense, these various interpretations of browsing share a central characteristic: they are scanning processes (Rice et al. 218).

In the final, refined framework for browsing (Rice et al. 2001: 295), which is a slightly modified version of the original tentative model presented on page 234, the process of browsing is seen to consist of 1) a behaviour: scanning; 2) motivation: a goal; 3) cognition: an object; and 4) resource: form. Rice et al.'s summary of these components of the model goes as follows:

Thus, in the new model for describing the browsing process, scanning is the salient component of the behavioural dimension. Goal is the central element of motivational dimension. Object is the major cognitive process. And form appears to be characteristic of the resource dimension. (Rice et al. 2001: 291-292).

Their perspective could be re-worded as follows: browsing consists of the behaviour of scanning, driven by the motivation of having a goal directed toward an object. When the object is an information resource, the influential variable about that object is its form.

Is browsing really like this? How could we characterize the act of browsing to distinguish it from all the other types of acts human beings commit? Is browsing really only the intentional scanning of an object?

Instead, I would argue, browsing consists of a series of glimpses, some glimpses leading to further, closer exploration of the thing(s) glimpsed and some not. In this view, browsing consists of numerous stops and starts, with some reading, or surveying, alternating with other actions, such as sampling and selecting. Rice et al. (2001) are clearly well aware of this pattern to browsing, as indicated by their discussion of identification of units for analysis on page 242. The empirical observations made of people browsing were analysed into a hierarchy of case, movement and episode. Each case was an observation of an individual's activities in a library, each movement reflected movement from one part of the library to another and each episode represented a distinct goal or object within a movement (pp. 238-242). "A new episode within a movement begins when either the goal or the object under consideration changes and it is distinguishable by a question similar to movement: 'I observed you do… what was your intent with that activity?'" (p. 242). Rice et al. (2001) are thus clearly aware of the segmented, or step-wise, nature of the act of browsing, but this awareness is not evident in the final model.

Further, they produce a taxonomy, which consists of nine types of browsing and one type of non-browsing (p. 286). (These types are distinguished largely by the intent or context of browsing, not by differences in the act of browsing itself.) Both the browsing and non-browsing members of the taxonomy involve scanning; thus the one distinctive behavioural characteristic of browsing that the authors identify— scanning— is not, after all, truly defining for browsing, because scanning can occur in non-browsing situations as well. What, then, uniquely defines the act of browsing?

Elsewhere, the authors come much closer to dissecting the distinctive characteristics of browsing. They note "The four distinct acts of browsing were derived from the observation of scanning as a non-verbal behavior" (p. 290. ). These four acts are defined as looking, identifying, selecting, and examining (p. 291. ). After their brief mention, these intriguing activities are not incorporated in the final model of browsing, however and are not addressed again in the book. Nor did a literature review on the concept of browsing for the years subsequent to their book produce any further developments on these elements.

These intriguing components of browsing, which indicate that browsing is a more complex and interesting behaviour than just scanning, are addressed insightfully; however, in a short paper by Kwasnik (1992) (cited, but apparently not drawn upon substantively, by Rice et al. 2001). Her results are based on just four people and were intended to precede a larger study. Two of the individuals were observed in a farmer's market and the other two used a mail-order catalogue of gifts, so browsing both among physical objects and among informational representations was included.

Kwasnik notes:

In our analysis we focused on the movement of people's attention from item to item or from representation to representation. Movement was signaled by physical movement (walking, page-turning, finger pointing and so forth) and by verbalizations that showed evidence of a shift in focus. The definition of unit of analysis proved to be a thorny problem. If browsing is a kind of movement among a set of well-connected nodes, then what is a 'node'?

and

We then developed the notion of a view. A view is what a person articulates as seeing at one time, that is, a span of attention. We have some good clues for operationally identifying views because the participants almost always labelled them (Kwasnik 1992: 194).

Kwasnik's definition of browsing, which follows below, is, in my view, less than ideal— more on a definition shortly— but her characterization of the activities of browsers is entirely accurate:

We defined browsing as movement in a connected space. In order to achieve this movement, people undertake certain actions: they shift their gaze, they alter their position, they skip over things, they glance at things briefly, from afar, or close up, they back up, they pause or stop and they respond to interesting phenomena (Kwasnik 1992: 195).

Kwasnik then goes on to identify six activities that play a role in browsing:

  1. Orienting to the environment— a persistent activity, not only one at the beginning.
  2. Place marking— "Marking a view for potential second consideration" (p. 195).
  3. Identification— "or recognition of potentially interesting or definitely not interesting items" (p. 195).
  4. Resolution of anomalies. "Consistent with the notion that browsers create structure and orient themselves as they go along is their pronounced effort to resolve anomalies, even, it seems, when the item being 'resolved' does not seem to be of great interest otherwise" (p. 195).
  5. Comparison. "These comparisons serve to orient, identify and solidify purposes and aims" (p. 195).
  6. Transitions. "The movement from one view to another is a transition" (p. 195). Transitions may be movement toward an object of interest, or away from one, when the browser has identified something and rejected it or exhausted its information potential.

The proposed definition

Here the activities described by Kwasnik are interpreted to be a part of a combined cognitive, motivational and behavioural pattern that can be presented as follows (my definition):

Browsing is the activity of engaging in a series of glimpses, each of which may or may not lead to closer examination of a (physical or represented) object, which examination may or may not lead to (physical and/or conceptual) acquisition of the object.

Though the above may work as a definition, that definition needs to be analysed even more closely to express fully what the act of browsing is about. Human beings rely very strongly on vision; it is our predominating sense. In browsing, both visual movement and physical movement, to a greater or lesser extent, typically come into play. We may examine something only visually, or with touch or other senses, such as smell or hearing. If we want to acquire the thing examined, we may do so by reading, listening, smelling, or other activities that allow us to take in some experience about the thing examined. That taking in may include actual physical acquisition of the object, or be limited to informationally absorbing the object in some way.

Let us apply this definition to a prototypical browsing situation, looking at magazines at an airport news stand. Based on experience, you know that the news stand could be a browsing-rich experience. You walk over and stand close enough to the shelves to read the headlines. You glimpse a section. You see a headline or picture that interests you. You pick up the magazine, i.e., you select it. You read a bit, i.e., you examine it and put it back. You glimpse something out of the corner of your eye. You turn your head and look at that magazine. No, not interesting after all. You now look in the other direction on the shelf. Ah, now that's really interesting! You pick up the magazine and read a little. You think, "I'm going to buy this one. I can read it on the plane," i.e., you acquire. You then either browse additional magazines or go to the check-out stand to buy your magazine.

You have glimpsed, selected and acquired. Not every glimpse led to selecting and not every selection led to acquiring. You could have left the news stand unable to find something you wanted to buy. You might have walked up to the news stand and discovered that all the magazines were in the language of the country you are travelling in and you do not read the language, so you would not even have selected anything. There are many combinations possible out of these elements. Further, you probably moved your body in a variety of ways during this process, often to help or support the visual or haptic examination of the magazines.

We need to examine more closely the second step in this model of browsing. The word selecting has been used. Indeed, in many cases, the thing that catches our eye is something that we can expect to be drawn to by our pre-existing interests. You like skiing and cooking; the magazine's headlines mention some words in one or the other of those areas, the words catch your eye and so you reach out and pick up the magazine to look at it more closely.

On occasion, however, the act of selecting should more properly be called sampling. For example, you walk up to a rack of blouses in a department store. The colors of the articles are not very appealing; as a consequence, you are not drawn to any one item. But in hopes that you might find something interesting after all, you sample one of the items by picking it up off the rack and looking at it more closely. Maybe there will be something cute in the design of the blouse after all. Here, you are truly sampling; one blouse, or one color, is as good as any other in order to see the styling.

So, sometimes, perhaps most of the time, we select something in particular to look at out of the field of vision created by the glimpse. We make this selection because a preliminary assessment of that field of vision has led us to feel the most attraction to or interest in that one particular object within the field of vision. We select that something because we prefer it to any other object in the field of vision. This second step in the browsing process is properly called selection. At other times, however, as discussed in the preceding paragraph, this second step is more properly called sampling, because we are more or less randomly reaching out to examine any one of the objects in the field of vision; it does not matter to us which one.

With these thoughts in mind, browsing can be seen to contain four elements, iterated indefinitely, until the overall episode ends:

  1. glimpsing a field of vision;
  2. selecting or sampling a physical or representational object from the field;
  3. examining the object; and
  4. physically or conceptually acquiring the examined object, or abandoning it.

At a minimum, all browsing consists in, at the least, glimpsing a field of vision, abandoning it and glimpsing again, iterated until the end of the browsing episode. Most browsing episodes, however, include repeated selecting and/or sampling and repeated instances of examining objects. Further, browsing episodes may contain one or more instances of acquiring examined objects, which acquisition may be conceptual, as with reading, or physical, as with picking up and keeping the item.

Another way to describe browsing is to say that it is multi-step: 1) glimpsing, that is, acquiring a field of vision, 2) latching on to an object within that field of vision, 3) examining that object, 4) keeping the object, or abandoning it. The next moment in a browsing episode begins with the first step, glimpsing.

At first, the work by O'Connor on browsing appears to come closest to the approach taken here. He, too, describes four phases, beginning with "Make Glimpses" (O'Connor 1993: 223). However, his emphasis throughout is quite different. He is primarily concerned with the cognitive processes involved in processing the information seen and less so with the act of browsing per se. His four phases are "Make Glimpses" (p. 223), "Connect Attributes" (p. 224), "Evaluate Connection" (p. 226) and "Evaluate Search" (p. 227).

Even his "make glimpses" is about picking a starting point for evaluating the content of a collection or browsed area. "The individual glimpse is the instrument which enables the searcher to create the appropriate representation system" (O'Connor 1993: 223). The searcher then determines the attributes (document features) that will be examined to determine if an item is of value or interest. The second phase, connecting attributes, involves matching the attributes of documents seen with the attributes wanted by the searcher, the third phase, evaluating connection, involves determining the degree of overlap between searcher and document features and the fourth phase, evaluate search, refers to assessing the value of the connections discovered.

Returning to the approach to browsing that has been presented here, when does browsing behaviour stop being browsing and become something else? For one, it can be said that scanning is not browsing. As noted earlier, scanning involves, just as Rice et al. (2001) said, a systematic, smooth movement. But browsing, as described here, is very different from that. Browsing is a complex process, involving a series of glimpses, usually followed by actions between the glimpses. Browsing is not a smooth scan. I glimpse one section of the magazine stand, seize something interesting within it, put it back, then glimpse again. This is anything but a case "where the person's body or eyes move smoothly at will" (Rice et al. 2001: 178).

There are, in fact, cases where we do scan or skim. Generally, we engage in these smooth, sequential, orderly activities when we want to find something quickly but want to be sure not to miss anything (American Heritage Dictionary's second sense of "scan"). For example, suppose we have looked up an encyclopedia article alphabetically and now are scanning the article from start to finish to locate a specific fact that we guess should be in the article. Browsing, by contrast, is more open to surprise. The glimpse may land us anywhere within a reasonably wide visual range. Scanning or skimming, on the other hand, are very focused. The reader is reviewing as fast as possible, but systematically, in order to cover everything within a chosen area and not miss anything that might be there. To return to the magazine stand, we seldom browse by starting at one end of the racks and smoothly scanning across every rack, one after the other, left to right in a systematic way, in order to be sure to see every magazine as we scan. Rather, we move around and our eyes move around, landing here and there on things that interest us. We may well miss many magazines in the process and glance at some repeatedly, in true browsing.

One final point: Browsing has been defined in terms of vision, with a series of glimpses seen as the sine qua non of the act of browsing. It seems possible in principle, however, to extend this definition to, say, browsing sound clips on one form or another of a music player. Further, people who are blind presumably also find some way to engage in this fundamental behaviour, through touch or other means.

Support for this approach from other disciplines

I cannot claim the Ph.D.'s expertise in the fields from which literature will be cited and discussed below. Consider this section speculative. However, nor is it the case that marginal authors are being sought out and emphasized from these other fields. I have some sense of the mainstream, or at least ideas widely recognized and taken seriously, as opposed to fringe approaches. Though I lack the professional qualifications to make the case fully, the ideas presented below provide strong preliminary justification for pursuing the line of attack proposed in this article.

Psychology: visual search

Very strong support for the visual aspects of the above model come from psychology. Wolfe's model of visual search (1994), called Guided Search, has been cited over 700 times and is still considered a valid contender among visual search theories at least as recently as Logan's 2003 review article on theories of attention (Logan 2003). Basically, Wolfe's model holds that we glimpse widely initially, then home in on points of interest within the broader visual field for more complex interpretation and understanding.

… the model distinguishes between a preattentive, massively parallel stage that processes information about basic visual features (color, motion, various depth cues, etc.) across large portions of the visual field and a subsequent limited-capacity stage that performs other, more complex operations (e.g., face recognition, reading, object identification) over a limited portion of the visual field. … . The heart of the guided search model is the idea that attentional deployment of limited resources is guided by the output of the earlier parallel processes (Wolfe 1994: 202).

Thus, in visual search, we first glimpse widely, processing only the simpler, grosser characteristics of the scene in front of us. This massively parallel stage means just that; a wide area is viewed all at once, not in a smooth scan from one side to the other. The output from this initial glimpse, namely detection of possible points of interest in the visual scene, leads to a second stage where a point or points of interest is processed more intensely and with more sophisticated mental capabilities, such as the ability to recognize faces, or to read language. These latter capabilities take lots of processing space in the brain. "There is not enough room in the skull for all of the neural hardware that would be required to perform all visual functions at all locations in the visual field at the same time… . A large set of visual functions can be performed only in a restricted part of the visual field at any one moment" (Wolfe 1994: 202).

… there are parallel processes that operate over large portions of the visual field at one time and there is a second set of limited-capacity processes that are restricted in their operation to a smaller portion of the visual field at one time. In order to cover all of the visual field, these limited-capacity processes must be deployed serially from location to location (Wolfe 1994: 203; italics are Wolfe's).

Thus, we glimpse a scene all at once, then select one or more things within the scene to examine more closely. We then examine the selected objects with more sophisticated, processing-intense capabilities. These latter capabilities use so much brain capacity that they can be applied only to selected, smaller parts of the initial visual field. If we want to examine more than one element of the scene with these powerful capabilities, we must examine the elements one after the other, i.e., serially.

Psychology and behavioural ecology: curiosity and exploratory behaviour

Modern discussions of curiosity begin with Berlyne's classic, Conflict, Arousal and Curiosity (Berlyne 1960). The frontispiece of the book is a photograph of a cat standing on a chair with its front paws perched on the top of a typewriter, sniffing at the typing mechanism of the machine. Curiosity and exploratory behaviour seem to be found generally in animals and, of course, cats are supposedly among the most curious of all, to the point of getting themselves killed on occasion.

For browsing, especially of texts, human reliance on visual search can be almost total. However, there is a kinesthetic element, too, in much browsing. After identifying something of interest, we may examine it only visually, but in many cases, we also move in closer to the item, pick it up, or otherwise move in some way in the process of examining it. With our nerve-rich fingers and opposable thumbs, haptic examination often becomes an important part of exploratory behaviour.

Interestingly, Berlyne posited three stages of exploratory behaviour, which remind one of the elements of Wolfe's visual search model. Berlyne's three elements were: 1) orienting responses, 2) locomotor exploration and 3) investigatory responses (1960: xi). There is a rough parallel here to 1) the initial wide glimpse, 2) homing in on some specific object or feature of the environment and 3) examining it.

It would appear that something very fundamental to the nature of animals, including human beings, is being described here. Just as was argued by the visual-search researchers, efficiency in processing may require this approach of surveying the territory first, followed by a focus on some part of the territory. After all, other kinds of search could be carried out by animals, such as a serial examination of one thing after another in the territory, with no preliminary survey. We may guess that evolution has led to survival of the most efficient exploratory method.

So why do animals explore? What model can psychologists use to explain this behaviour? In a recent review, Hughes (1997) traces the history of the various theories developed to explain exploratory behaviour in psychology. Hughes defines intrinsic exploration as follows: "Intrinsic exploration involves exploratory acts that are not instrumental in achieving any particular goal other than performance of the acts themselves" (1997: 213). This is contrasted with extrinsic search, which is driven by some goal, such as the need for food or escape from danger.

I will jump to the bottom line by saying that Hughes concludes that there is no persuasive model at the time of his review to explain this behaviour. He notes that animals seem to manage to stay between extremes of boredom on the one hand and over-stimulation, on the other. Hughes concludes, "If a motivational force has to be proposed, it is probably sufficient to go no further than accept that… organisms appear to have some type of behavioural 'need' for sensory change which can be satisfied by intrinsic exploratory responses" (1997: 219).

Loewenstein, in an extensive review of curiosity (1994), likewise found available theories wanting and only partially explanatory. He posits a sense-making theory for explaining curiosity.

… the information-gap theory views curiosity as arising when attention becomes focused on a gap in one's knowledge. Such information gaps produce the feeling of deprivation labelled curiosity. The curious individual is motivated to obtain the missing information to reduce or eliminate the feeling of deprivation (Loewenstein 1994: 87).

This description comes perilously close to being a circular explanation. We discern a gap and want to fill the gap. But why do we want to fill the gap? Why does having this gap make us feel deprived? Does not the word gap already imply a deficiency? Perhaps this is just a space in our knowledge. Why should this space in our knowledge lead to curiosity? As Loewenstein says on the same page: "As noted earlier, the remaining question— the cause of curiosity— is inherently unanswerable" (p. 87).

However, it seems that one can posit a straightforward evolutionary explanation for curiosity, an explanation that makes this behaviour quite understandable. In motile (as opposed to sessile) animals, exposure to new environments or new stimuli or new information all bring with them the possibility of discovering new food sources, new mates, new nesting or sleeping sites, or new ways to escape predation. Thus, the ability to move, combined with the ability to sense the environment, had a positive payoff for the animal with these capabilities. Bell's massive 300-plus-page review of searching behaviour of animals throughout the animal kingdom, "The behavioural ecology of finding resources," as his sub-title states, finds that exploratory behaviour pays off for animals in a variety of ways (Bell 1991: 165-169). However, the brevity of his chapter on exploratory behaviour (five pages) within the larger body of research on animal searching confirms his statement therein that "Exploratory behaviour is probably a more significant component of searching than is usually appreciated" (p. 165).

Let us develop this idea more. Exploratory behaviour in any territory previously unknown, whether of the physical world, or, in humans, of emotional or intellectual territory, can lead both to new, valuable discoveries and to devastating harm. Tasting a new mushroom found in the forest, slipping off the edge of a heretofore unseen crevasse, being captivated by a charismatic but dangerous leader, or fretting in a sophomore philosophy class about the existence of free will, can all lead to damage or even death for the vulnerable individual.

There is an inherent trade-off in exploratory behaviour, between the good things that can come of it and the danger that can result from it. Over the evolutionary history of this planet, this trade-off must have been played out a googolplex (more or less) of times. For each currently existing species, we can expect that a tendency to explore and a tendency to stay put or traverse only familiar territory exist in the current members of that species in a balance that is a result of natural selection. Too static a pattern and the animal may lose out to competing species that explore more and discover more things of value to their survival. Too much exploration in risky environments and members of the species may die or fail to reproduce often enough and thus harm the survival of the species. After millennia of natural selection, what remains is a balance between the impulse to explore the unknown and the impulse to stay with the familiar. This balance represents the cumulative trade-off between these risks that the species has encountered in its history.

Cats may or may not have a higher level of curiosity than do human or other animals. But suppose, for the sake of argument, that they do explore more than many other animals. They probably have this level of exploration because, over their evolutionary history, in the environments in which they existed, that level of curiosity paid off positively for cats. Another animal, in a different environmental mix, might have been harmed by being so curious and so came to explore less.

General exploratory behaviour in humans is manifested in a number of ways, with many of the activities being similar, though not necessarily identical to browsing information: shopping, Web- or channel-surfing, mingling at parties, sightseeing (sampling new experiences), dating, nibbling at a buffet table, etc. (Jenna Hartel 2003, personal communication). All these activities may be manifesting a general exploratory impulse, with the physical and psychological constraints of the particular situation shaping the specifics of the cognition and movement.

Berrypicking

This author has written on a related behaviour labelled berrypicking, and the distinction between berrypicking and browsing should be made clear here. Berrypicking characterizes whole information searches, or, more precisely, whole episodes of information searching. A berrypicking search has an evolving query, which shifts during the course of the search. As the searcher discovers sources and learns new information, the query changes accordingly; the searcher adds and drops elements and sometimes re-conceptualizes the whole query as understanding of the relevant information domain increases. The searching during the episode is in a berrypicking mode, that is, involves getting a bit of information here, another bit there, just like picking berries in a forest. In a berrypicking search, there is no assumption that the system will produce a single complete, final retrieved set. Finally, during the course of an episode, the berrypicking searcher may use many different sources and may use varying searching techniques, each suitable to the source and circumstances of the search.

Thus, berrypicking characterizes whole information searches or search episodes. When browsing is done in the course of an information search, it may play a part in many different types of search strategies and techniques. The searcher may adopt browsing and then cease browsing seconds later in the course of a complex berrypicking search.

Here is an example:

The searcher is doing a berrypicking search on information seeking behaviour among humanities scholars. She has done a directed search for the topic in article databases, searched by title keyword in library catalogues and read over the reference lists in several books and articles she has found to this point in her search. These approaches involve several different search techniques and she has collected relevant material, like picking huckleberries in a forest, here and there, bit by bit. As she goes, she gets a better sense of her topic and narrows and changes her focus (evolving search). While in the stacks finding the books she liked from the catalogue search, she browses among the neighbouring books to see if any look promising. She pulls out some of the books that catch her eye, randomly browsing a bit on this or that page. Something she reads in one of the books reminds her of an article she has seen recently. What was that author's name? Started with a W. She goes back to an article database, searches by topic and then scans down the list of retrievals, looking for last names starting with W. Ah hah! There it is. She prints off the article, then scans down the list of references in that article to see if there is anything good that she has missed… .

Scanning, browsing, directed searching were all involved in this berrypicking search.

Implications for information system design

If the cognitive/behavioural act of browsing is different from what Rice, et al. (2001) assume and closer to what Kwasnik (1992) identified and I have conceptualized and if, as I have argued, it arises from a fundamental animal exploratory behaviour that goes back millions of years and has long rewarded animals (and also put them at risk), so what? What are the implications for information systems?

Browsing appears to be a manifestation of a fundamental animal exploratory behaviour. If so, it is natural to people and can be engaged in spontaneously, without training, provided information system interfaces lend themselves to this behaviour. Most browsing capabilities in Web-based and other online information systems consist of the capability of opening some text or images and scanning down a long list or a set of thumbnails. Such a capability is better than nothing, but it does not facilitate browsing very well. Put differently, such design facilitates scanning, but not browsing.

These designs appear to be based on an assumption that browsing equals scanning, rather than being based, as argued here, on a deeper understanding of the nature of browsing. Good browsable interfaces would consist of rich scenes, full of potential objects of interest, that the eye can take in at once (massively parallel processing), then select items within the scene to give closer attention to.

Elaine Toms's (2000) research on browsing is very revealing in this regard. She studied people browsing articles from newspapers, presented in an interface in such a way that the searcher could see article titles and go off and review different articles at will. Thus, the people browsing were not scanning so much as clicking on points of interest that caught their attention, and a click would lead to an article.

Specifically, in addition to a window in which to view the actual article text, there were three key elements to the interface: menus of broad topics, a conventional search tool requiring a term to be input and a suggestions list of related articles. (Items on the latter list were automatically generated by algorithms that identified similar articles.) In a typical screen shot in Toms's study, half the screen is taken up by a window displaying the text of the latest article being examined. (See Toms 2000: 430-431, Figures 1 and 2) The other half contains two windows, one displaying menus and the other displaying either the Search Tool or the Suggestions. Toms noted the following in her discussion:

Overall, participants were quite clear about the function of the different tools. Menus were seen as a structured approach to examining the newspaper and were widely used throughout the experiment… . The menu was treated as an anchor in the interaction with articles and served for route-finding and orientation. For example, not once did participants comment about being disoriented or getting lost, typical reactions to many information systems… . The Search Tool was acknowledged as essential for looking for information on a specific article, while the Suggestions provided the diversion and encouraged meandering which then uncovered items of interest to participants. Participants clearly understood these roles (Toms 2000: 446).

I argue that the design used by Toms better supports the natural physical sense of browsing that feels most native to human beings, than do many other experimental and operational systems. In effect, the menus provided the orientation that is normally implicit in the physical environment, but which must be provided explicitly in some way in an electronic environment where screen can instantly replace screen and thus disorient the searcher, as in a typical World Wide Web browser. In the design she used, Toms implicitly demonstrated that orientation can be retained without sacrificing the sense of movement implicit in the clicking and going to another document. Because moving around from screen to screen left browsers disoriented in some past experiments, designers may have cut back on the movement and limited the options for the browsing individual mostly to scanning. Instead, in this design, Toms allows the browser to remain oriented through use of the menus, while glimpsing the Suggestions and clicking on them (moving to them) to examine them more closely.

This glimpsing and moving to another screen permits the browsing process to mimic much more closely the physical act of browsing that takes place away from the electronic world. The browser takes in a number of article titles at once (ten titles appear in a list in the upper left corner of the screen, taking up about a sixth of the whole screen area; see her Figure 1, p.430). Clicking on one title is equivalent to moving to an item and picking it up or looking at it more closely in the physical world. The full article text appears in the screen text window while the suggestions and menus remain in other parts of the screen, thus enabling the searcher to remain oriented (cf. Kwasnik 1992). Here the browser can acquire, i.e., read, the article, or just glance at it and move on. A return to the menu or suggestions list is equivalent to putting the item down in the physical world and continuing to browse.

Toms was not testing the model presented herein, of course. Her definition of browsing was "an activity in which one gathers information while scanning an information space without an explicit objective" (Toms 2000: 424). But intentionally or not, the online system she tested supported or paralleled the physical behaviour of browsing in the real world better than do many other online information systems. I suggest that in her research, she implicitly demonstrated the superiority of the model of browsing provided herein. Obviously, a more explicit and self-conscious test of this model would be needed for greater confidence that my model is a good one. One can reasonably suggest, however, that testing the model of browsing presented herein, both in physical and electronic environments, is a very promising direction to go in and may lead to a more accurate representation and understanding of the human behaviour of browsing than we have had to date.

Summary and Conclusions

Based on the discussion herein, what can we say about browsing?

  1. Animals that can move (motile as opposed to sessile animals) may use their movement capability to explore in a more or less random manner. This behaviour exposes them to new places and possibilities. Such exposure may lead to new food sources, mates, nesting or hiding places and escape routes from predation. Such exposure may also lead to harm or death.
  2. Presumably, the amount and extent of exploratory behaviour exhibited in a given species is the result of the historical trade-off experienced by the species between explorations that led to positive results and explorations that led to deleterious results. (Another, non-conflicting, possibility is that there may be a range of exploratory behaviour within the species as well, with some animals doing more or less exploration. Thus, this variation is the engine for further natural selection and, therefore, evolution of the species' exploratory propensities in one direction or another.)
  3. Based on the above, we accept that most animals have a propensity toward exploratory behaviour.
  4. In humans, at least, visual search is optimized for efficiency and constrained by brain capacity, in such a way as to require a particular kind of search. Specifically, we take in a scene all at once in a massively parallel glimpse, then select or sample a spot within the glimpsed area to examine more closely, using higher-level capabilities which require much more mental processing space. The glimpse notes basic features such as colour and movement. The higher-level processing at the examination stage engages in such activities as face recognition, reading and object identification.
  5. Browsing is a cognitive and behavioural expression of this exploratory behaviour. The in-built motivation for this exploratory behaviour can be called curiosity. Because humans are so strongly reliant on vision, bodily motion often mirrors visual search, in that the second stage of browsing often involves physical movement toward items of interest, which movement, of course, also supports closer visual inspection.
  6. Browsing is here considered to have four elements. The first element is essential to our understanding of browsing, the later elements almost always occur as well: 1) glimpse a scene, 2) home in on an element of a scene visually and/or physically (if two or more elements are of interest, they are examined serially, not in parallel), 3) examine item(s) of interest, 4) physically or conceptually acquire or abandon examined item(s). This sequence is repeated indefinitely through further glimpses. Browsing is thus not a smooth scan of a scene.
  7. Formally, browsing is defined thus: Browsing is the activity of engaging in a series of glimpses, each of which exposes the browser to objects of potential interest; depending on interest, the browser may or may not examine more closely one or more of the (physical or represented) objects; this examination, depending on interest, may or may not lead the browser to (physically or conceptually) acquire the object.
  8. The design of interactive information systems needs to incorporate an awareness of human browsing characteristics. Specifically, browsing for information in such systems should not be limited to the opportunity to scan, but instead enable the searcher to manifest the instinctive tendency to engage in a browsing sequence: to glimpse, then to examine or not something glimpsed, then to keep or not the things examined.
References

How to cite this paper

Bates, Marcia J. (2007). "What is browsing—really? A model drawing from behavioural science research" Information Research, 12(4) paper 330. [Available at http://InformationR.net/ir/12-4/paper330.html]
Find other papers on this subject



Bookmark This Page

counter
Web Counter
© the author, 2007.
Last updated: 15 September, 2007
Valid XHTML 1.0!