Information Need: A Theory Connecting Information Search to Knowledge Formation

The first part of this book investigates what is meant by information need. Chapter 2, on the history of information, considers the work of early theorists such as Shannon, Paisley, Allen's formal and informal channels of information flow in information search, and integrates Allen's six channel types with Shannon's information flow channels. The next chapter introduces one of the main components to the theory developed in this book. The framework used is Taylor's four levels (Q1, actual, visceral need; Q2, conscious within brain description of the need; Q3, the formalised statement of the need; and Q4, the compromised need in language and syntax that the user believes is necessary for the information system). Chapter 4 contrasts the information science and computer science perspective on the user. According to this, using the Shannon model of information flow ideas, the information science perspective focuses on the exploratory search situation with high information content, which is modelled as four equally probable messages (each of 25 per cent probability) with the computer science perspective, with low information content, where one member of the four member set of possible messages has 90 per cent probability of being selected by the receiver –the question‐answer factual response situation. Charles Cole does not go into further detail about the computer science perspective, but it would be helpful to have had some examples from the literature of the type of computer science approach that he had in mind. For example, in a preface to a 2004 issue of the User Modeling and User‐Adaptive Interaction journal Brusilovsky and Tasso (2004) consider four papers on applying the ideas of user modelling for four different types of information access: ad hoc information retrieval, information filtering, hypertext browsing, and information visualization. The web has spawned more research on adaptive information access that blends some of the previously independent paradigms of retrieval, filtering, browsing and visualization. Cole's view on the computer science perspective seems to fit the idea of the information retrieval system that answers the user's information need expressed as a formal query, but this seems a very narrow definition given the research that has been conducted on adaptive information access, for example.

Chapter 5 sets out the premise that we should assume information need is unknowable, and that surrogate concepts of information need, information and information use need to be considered to develop a theory of information need. This does not seem intuitively useful, but Chapter 6 explains that information and information use are closely associated in the book, for the purpose of considering information search. The theories used to discuss perception (of messages from the information system) and state of readiness for the stimulus received, include Minsky's frame theory of perception‐thinking‐reasoning and (in Chapter 7) Harnad's categorical perception theory. Both of these theories appear to fit the theory of information need that Cole is developing, but it would be desirable to indicate that cognitive psychology is a huge area of research. Believe me, the Handbook of Categorization in Cognitive Science is a much more weighty volume than might be assumed from reading of Cole's chapters on the topic (Cohen and Lefebvre, 2005). Harnad, in Chapter 1 of the Handbook defines categorization as any systematic differential interaction between an autonomous, adaptive sensorimotor system and its world. Using Harnad's definitions (and as applied by Cole), categorization comes before concepts (which must have those used to grounded theory standing on their head). For Harnad, cognition is categorisation, and categorisation is essentially all about abstraction – we detect features, and differentially weight them, we single out some subset of the sensory input and ignore the rest. But this is only one viewpoint. The work of Minsky is invoked to account for the networks of concepts used in perception and reasoning. I wondered whether a similar approach, that of scripts or scenarios might not be more useful for some types of information search. Frames are static representations, whereas scripts (as described by Schank and Abelson) concern patterns of events. The ideas around scenarios are used by computer scientists for business process modelling, and use case modelling, but perhaps Cole considered this too prescriptive for exploratory search.

Chapter 6 explains how exploratory, pre‐focus information search should be modelled as positive feedback, to amplify the perturbing incoming signal. The user's expectation set, the conditional set of readiness has to be readjusted to provide a new explanation set to direct future information searching. This is described as “maximisation of perturbation via knowledge formation”. This is explained and justified in Chapter 7, using a circle framework. The unexpected environmental stimulus leads to a gap in Circle 2 (categorisation‐conceptualisation). Circle 2 represents the start of the perturbation stage, which is influenced by information flow around and from Circle 3 (associationism: the relational structure other object/event frames stored in memory), Circle 4 (belief system: values) and Circle 5, which relates to previous ideas discussed by Cole and Spink on evolutionary psychology. The basic problem is a devising a theoretical model to describe how the information searcher refocuses, revises original beliefs and tries to test out new ways of looking at the problem – how are these changes in what is perceived come about so quickly and how is existing knowledge supplemented and/or revised? It is, I admit, extremely ingenious but some of the theories on which it is based may be no more than untestable hypotheses. Donald's ideas about the development of the human brain come into this category. I wonder whether it is necessary to make this model so complicated. In many ways I would prefer a model that was related more closely to more accepted and testable ideas from cognitive psychology. Despite the focus of much cognitive psychology research on learning, artificial intelligence, text mining, and computational modelling with neuroscience, there is some research that offers approaches that do not depend on untestable claims. Surely there are ideas from research on learning, and models for learning that could be applied to the ideas about the perturbation stage? For example, Gärdenfors (2005) proposes an approach that brings together ideas on induction, and learning through connectionist systems (networks) in the form of “conceptual spaces”. These have different dimensions, but the basic idea is that the two objects near each other in conceptual space along one dimension are more similar to each other, than two objects that are farther apart. We may have dimensions that are culturally dependent (and this appears to be important for the Cole model). The example cited by Gärdenfors (2005) is the concept of APPLE, which is correlated to sweetness in the taste domain, sugar content in the nutrition domain and a weaker correlation between the colour red and a sweet taste. The conceptual spaces representation does get very mathematical with sets and Voronoi partitioning, but this representation allows for changes in our ideas about how we should perceive a category (shifting the prototype slightly). It also allows for the way in which different contexts in which a concept is used may trigger different associations, and this helps the ideas around the perturbation stage when we change focus. But this is only one suggestion, and the debate about categorisation and concept formation among cognitive psychologists continues (Machary, 2009).

The final part of Part 1 of the book proposes a theory of information need that integrates Taylor's four levels of information need, integrates Bates' ideas on berrypicking with the Kuhlthau six stage Information Search Process model, and also uses some task‐based frames (based on Belkin and previous research by Cole). I shall refrain from trying to describe the diagrams – you have to buy the book for that. Instead, the six propositions are set out, in outline. The first states that there are different types of searches, some of which have a command (known item type) intention, and some which have a question intention. However, both types of searches are motivated by the four levels of information need (as expressed by Taylor). The second states that it is not the information need that changes over the exploratory stage of information seeking, only the aspects of the topic that the user selects for investigation. The third states that information use evolves as the user's information seeking progresses, with the related fourth proposition stating that the information system envisaged is a knowledge retrieval/formation system which facilitates the progress towards the post‐focus stage when the Q4 level of Taylor can be linked to the deeper Q1‐Q3 levels and effective command searches are possible. The fifth proposition states that the socially imposed (task‐based) frames do not hook down to the Q1 (unconscious) information need, only as far as Q2 (conscious description of need). The sixth proposition states that information need at its deepest level is primarily a human adaptive mechanism, and thus Q1 at its deepest level touches on survival needs, and existential needs for seeking meaning.

Part 2 of the book explains with references to some examples from previous research (by the author) on how information need in the pre‐focus and focusing stages might be visualised. This requires (Chapters 10 and 11) definitions of types of known and unknown item searching and how the associative network would operate. Chapter 12 goes on to discuss the process of selection, including discrimination and identification. The discussion uses the idea of an association wheel, a constantly shifting and evolving process during information search. Chapter 14 discusses focusing search, involving Circles 4 and 5 (belief systems and values, mythic culture). Chapter 15 gives an example from research on history PhD students, Chapter 16 considers undergraduate history and psychology students searching for information for essay writing. Part 3 proposes an information system design that would help searchers, particularly in the uncomfortable Stage 3 of focusing.

Is the theory coherent? Would it stand up in other settings and for other types of information search that are not educational in focus? According to Merton (1949), “Middle‐range theory involves abstractions, of course, but they are close enough to observed data to be incorporated in propositions that permit empirical testing”. Returning to the six propositions, number one could be tested against identification of different types of searches. For example, could the type of searches described by Bowler (2011) be explained by this model? Proposition number two seems to me to be an article of faith, or a standpoint. Proposition three (with proposition four) are testable. Much of the development of the theory seems to depend on the thoughts and actions of the individual information searcher, and collaboration, discussion or interaction with other information resources, including people, seems to be excluded. Propositions three and four should be set against other types of user experience for searching. With the fifth proposition we seem to be back to standpoint, and proposition six is not, in my opinion, based on a sound evidence base. It is an interesting idea, and worth some consideration, but it is hard to see how the proposition could be tested.

One test is to check how well the theory stands up against some recent literature. The work by Wilson et al. (2009) on the evaluation of advanced search interfaces is another attempt at integrating two models (Belkin and Bates) to evaluate searching interfaces for the support of more explorative, ill‐defined information seeking. Three faceted browsers were investigated. This is a much narrower perspective, with the focus on the individual searcher, and their interaction with particular browsers. The emphasis is on the support of tactics, which makes the evaluation much easier to conduct. It is computer science (in some respects) but it most certainly is not simply the command type search that Cole seems to ascribe to the computer science perspective. This research does share some of the aims that Cole has for information systems that provide for the idea tactics and term tactics identified by Bates in 1979. The information need is not the focus for Wilson, but there should be some links between the two approaches. I suspect that the full implementation of the Astrolabe information system outlined in Chapter 18 would benefit from the details of tactics tested in Wilson et al. (2009).

For the information literacy audience the book offers valuable ideas. Information need is of course more than the background requirement for searching skills or putting together a search strategy. The books discusses how perception works, how shifts in perspective often need to occur, how different connections need to be activated – all these ideas are necessary to get away from an emphasis on devising neat search strategies and instant appraisal of the retrieval outputs. Sadly, much of student searching is often quite instrumental and I could be very cynical but for many students the unchanging information need may be to pass the assignment. However, meeting the learning objectives may indeed require some deeper thinking about the framing of the question, and the transformation required to think more deeply about, and around the topic.

I was impressed by the depth of scholarship evident in this book. I admired the honest, and coherent attempt to integrate existing research and theories. It might have been good to know what had been considered and found wanting, but that is probably asking too much! I do not agree with some of the arguments and propositions but that is not the point. There is something to argue with – that is the point.