Social Phenomenology: Implications for Information Retrieval Systems

Abstract

This paper explores the shift in focus of the information sciences from cognitivist theories of information behavior to the everyday information practices of individuals. By “everyday information practices,” Savolainen (2008) means the processes of information seeking, information use, and information sharing. This shift emphasizes the conscious, lived experience of individuals in their discrete informational environments. Whereas previous studies of information behavior revolved around objective analyses of information systems, the new science attempts to qualify the subjective phenomena of information users. These phenomena include user perception and affective mental states, which have become the main focal points in attempts to study human-computer and human-system interaction. This empirical approach to studying information behavior has the potential to change how information is delivered and presented to users in various contexts. Following the social phenomenology of Savolainen (2008), I argue that the concept of lifeworld and users’ unique psychological characteristics are a vital consideration when evaluating or designing an information retrieval system.

Introduction

Understanding the psychology of information users is important in the digital age, as the need for information today is often punctuated by high demand and fast-paced work tasks. This means that information users are often busy in their daily lives performing retrieval tasks with the ubiquity of computer-mediated technologies. Information users, however, can easily become absorbed in these technologies, causing them to miss out on important opportunities for socialization. Recently, Kushlev et al. (2017) studied information-seeking behavior using smartphones to determine the effects on social and emotional well-being. Using experimental methods, the authors found that individuals who used smartphones for everyday information tasks felt more socially isolated than their “phoneless” peers. Because social connection is an essential psychological need, Kushlev et al. surmised that pervasive technological information seeking and use can have a negative impact on affective mood states (p. 72).

Social phenomenology confers the same understanding, allowing that affective mental states can radically influence user information practice. By “everyday information practices,” Savolainen (2008) means the processes of information seeking, information use, and information sharing. Each of these activities depart from one’s lifeworld, or the totality of experiences comprised in every individual (p. 51). This totality concerns the subjective viewpoints of individuals, the interplay of these viewpoints with others – referred to in the phenomenological tradition as intersubjectivity – and the social environments where individuals are embedded. Because users tend to assimilate new information based on what they already know, subjectivity dominates the information-seeking and retrieval process. Subjectivity is also greatly influenced by feelings and emotions, which are ingrained in one’s lifeworld. Cognitive and information scientists have gradually come to accept that affective mental states are precognitive and that they invariably influence cognitive behavior. Because cognition is required for information retrieval, it is important to consider the social phenomenological perspective when evaluating or designing an information retrieval system.

Literature Review:

The literature on information behavior over the past thirty years has increasingly considered the psychology of individuals. Beginning with Kuhlthau (1991), the process of searching for information was considered from the user perspective. Drawing on the work of psychologists and information scientists, Kuhlthau recognized that information needs were often preceded by uncertainty or anxiety. In order to ameliorate this anxiety, users had to distill meaning from information they retrieved during the information search process. Kuhlthau designed a model called the Information Search Process (ISP) model, which incorporated the affective (feelings), the cognitive (thoughts), and the physical (actions). The model also described six stages of information seeking. These are task initiation, topic selection, prefocus exploration, focus formulation, information collection and search closure (pp. 366-68). Kuhlthau’s model was applied in a large-scale, longitudinal study which surveyed 385 library users at 21 institutions including academic, public, and school libraries (p. 365).

Essentially, Kuhlthau argued that there are causal associations between cognitive states and affective states. For example, initiation of a new research task for high school or college students – the principal participants in Kuhlthau’s study – will cause uncertainty (a cognitive state), leading to anxiety and a lack of confidence (affective states). But as students embark on the research process and gain more knowledge of their topic, their confidence will increase. However, a dip in confidence during the exploration stage of the ISP is inevitable, as students will feel overwhelmed by the amount of information on their selected topic. This will cause them to revert back to a negative affective state. As students begin to identify relevant sources and interpret documents, there will again be a concomitant rise in positive affective states. Broadly understood, Kuhlthau’s ISP model indicates that higher knowledge states directly correlate to more positive affective states.

Kuhlthau’s model was widely accepted and was the first model of the research process to include emotions as an important factor for research. The ISP model was subsequently used to teach students to accept negative feelings as a natural part of the research process. This idea was embodied in Kracker’s (2002) opinion that “correctly placing the root of the negative emotions within the [research] process itself, rather than within the individual, allows students to experience the emotions in a less threatening and less personal way, and may help them feel less responsible for the discomfort they encounter” (p. 290). Recognizing that anxiety and stress can interfere with learning, Kracker devised a 30-minute presentation of Kuhlthau’s study for a group of high school and college students. Results suggested that knowledge of the ISP model reduced anxiety, but the study fell short in determining how the end product of the research process was perceived both emotionally and cognitively. Therefore, in another study, Kracker and Wang (2002) used content analysis to survey the psychological impact of research. The authors surveyed ninety students from four different sections of a technical and professional writing course (p. 296). Analyzing these students’ natural language descriptions of their own personal research projects, the authors coded 122 words representing feelings and 25 affective categories that group words based on similarity of meaning (p. 298). The affective categories measured emotional states, perceptions, and affinities toward the research process, thus encompassing both positive and negative feelings. 8 cognitive categories were selected based on Kuhlthau’s ISP model, comprising Kuhlthau’s six stages of information seeking, as well as “overall view of research” and “iterative nature of the process.” In addition, 5 derived categories were added measuring students’ thoughts on writing, learning, creativity, assistance from others, and locating sources (p. 299).

Kracker and Wang searched for co-occurrences between the two phenomenal categories of affective and cognitive. For example, feelings of anxiety were reported for 10 different cognitive categories (p. 300). However, feelings of confidence were reported for all of the same categories except iterative nature and writing process (p. 300). While more students indicated anxiety over confidence (69 mentions versus 31 mentions, respectively), the number of mentions across Kracker and Wang’s study were widely scattered. For example, discrepancy existed between feelings of difficulty and easiness (48 mentions versus 24 mentions), and feelings of dislike and interest (15 mentions versus 32 mentions) (p. 300). This last finding is odd, as it seems that positive affinity toward research should be low compared to the relatively high mentions of negative mood states. Also notable is the fact that Kracker and Wang found elective students as having had a more negative experience than major students (p. 300). Explications for these findings depended on further study, which problematized the authors efforts of finding patterns and trends relating to student perception and affinity toward research tasks.

Despite the importance of this research in bringing to light the centrality of user perception and emotion, there are problems with the ISP model. Savolainen (2015) argues that Kuhlthau “does not characterize the relationship between short-lived affective factors (feelings) and moods (relatively enduring affective-cognitive factors understood in terms of attitudes)” (p. 184). Savolainen agrees that affect and cognition are interrelated, but instead of defining mood as an orientation or an affective mental state that can be, as it were, conjured at will – as with Kuhlthau’s invitational and indicative moods (1991, p. 363) – Savolainen’s definition of mood takes on a more complex, phenomenological meaning. What Savolainen finds at fault is the functionalism inherent in Kuhlthau’s research. Writing further, Savolainen says: “Following the ideas of [George] Kelly (1963, p. 64), mood has been assigned a more central role in the ISP model because it allows the range within which the cognitive factors can operate to broaden or narrow (p. 184). In other words, it seemingly did not matter to Kuhlthau what constituted mood; only that mood methodologically supported the processes of the ISP. This framework presupposes that emotions do not come into play until they are triggered by the research process. Social phenomenology, on the other hand, employs a temporal dialectic where human action is rooted biographically and perceived as project-oriented and perspectivist (Savolainen, 2008, p. 53). This implies that the past can influence the future, and mood can influence information-seeking and use.

Nahl (2007) is another author who has advocated a user-centered focus based on affective processes. Nahl proposed the Social-Biological Information Technology (SBIT) model. In this model, there are three biological subsystems that operate during the stages of information processing and use. These are the affective, cognitive, and sensorimotor subsystems, which are believed by Nahl to operate interdependently. Briefly explained, the affective subsystem relies on a user’s emotions and motivations to use a system for information retrieval. When motivated, the cognitive subsystem analyzes the information retrieved to determine what to do with it in terms of “goal-directed planning” (p. 2025). The efficacy of this planning, however, is dependent on the intentionality of the user. For instance, the user may intend to use a system determinedly or, perhaps, apathetically. As Nahl says, “unless the user constrains thinking within the limits that count as goal-planning, the procedure does not count as planning, but as distraction, misinterpretation, serendipity, or shift in goal-intention” (p. 2025). Finally, the sensorimotor subsystem relies on the “use affordances” (p. 2025), or the design features that solicit user interaction with an information system interface. All three subsystems operate within the social context of the user and through the technological infrastructure of the user’s informational environment.

In the above description of the SBIT model, the adjective of constraint presupposes that emotions can be easily controlled by users in their efforts to direct the information search process. Aside from the initial affective motivations and intentions, Nahl does not see emotions as entering into the equation until the user has cognitively interpreted or assessed the information retrieved. Savolainen (2015) argues that Nahl’s framework hinges on appraisal theory, “suggesting that emotions are elicited by the individual’s cognitive interpretation or assessment of perceived information about the environment” (p. 188). In fact, as Savolainen shows, affective processes are the last step in the sequential procedures of the SBIT. For example, the sensorimotor subsystem notices information (step 1); the cognitive subsystem appraises information (step 2); and finally, the affective subsystem evaluates information based on feelings and emotions (step 3). According to Nahl, step 3 is the “end point” of information reception (Savolainen, 2015, p. 190). But what about emotions influencing cognitive processes before the search process even begins?

Outside of information behavior research, there have been some convincing studies in the field of cognitive neuroscience. Cohen et al. (2016) tested to determine how sustained emotional states impact cognitive control. The authors found that positive emotional states activated neuronal circuitry in the frontoparietal and frontostriatal lobes of the brain (p. 453). These are the areas associated with cognitive performance. When these brain areas were lit up on functional magnetic resonance imaging (fMRI) scans, individuals were more focused, and their cognitive skills increased. Conversely, diminished neuronal activity in these areas of the brain led to decreased cognitive performance, triggered by negative emotional cues (p. 454). However, like any cognitive study focusing on subjective phenomena, Cohen et al. were unable to empirically prove how their test subjects were feeling when they were submitted to positive and negative emotional cues. But the authors explicitly suggest that emotional states will invariably influence cognitive and neural processes.

It is more than likely that there is an additive effect between positive and negative emotional states and the development of one’s cognitive abilities. Here, the concept of lifeworld is useful. As stated above, users assimilate new information based on what they already know. In other words, the stock of knowledge for information users is sedimented (Savolainen, 2008, p. 57). We can utilize Savolainen’s idea of sedimentation when thinking about emotions as well. Throughout the course of lived experience, for example, emotions are compounded as individuals negotiate the lifeworld. These emotions can be predominately positive or negative, shaping an individual and their affective character. In developmental psychology, it is understood that this is how psychopathologies form (Cohen et al., pp. 446, 455). In light of the evidence on cognitive control, then, negative affective states are a vital consideration for information scientists. Especially considering that “one in five adolescents [meet] the criteria for mental illness” (Cohen et al., p. 455). This fact carries considerable import for post-secondary educators, but also for information system designers as well. Indeed, emotions are not – as Nahl suggests – spontaneous feelings that users experience only after appraising information. Nor are they moods that information users can solicit at will – as per Kuhlthau – to direct the research process. Instead, emotions are highly influential features that are rooted biologically and biographically in terms of the user’s lifeworld. Therefore, understanding how to deflect or redirect the negative emotions of users during the information retrieval process is an imperative for system design.

Discussion

Searching for information can be an enjoyable and enlightening experience, especially if that search is predicated on a user’s interests. Savolainen (2008) defines interest as a “teleoaffective structure” which “gives a general direction to thinking and provides a horizon for action (p. 57). But even personal interests can be adversely affected by negative mental states. Research interests in particular, and the cognitive effort it takes to perform research tasks, can be significantly influenced by negative emotions that are biological and biographical, but also stress-induced by the demands and time limitations of an assigned task. For instance, one of the first stages of the research process is resource gathering. This procedure relies on formulating search queries. As Savolainen (2015) says, “people may plan a search query by thinking of words they know that fall within the desired or intended topic” (p. 190). But this can be challenging for some users. Information users who suffer from major depression, for example, may have a harder time determining keyword relevance and synthesizing sources. Depression causes a generalized cognitive slowing and a deficit in word retrieval from memory. This deficit is due to “changes in the functioning of neural networks that coordinate complex cognitive abilities” (Douglas et al., 2012, p. 281). This is not to say that depressed users are poor researchers, or that they are incapable of producing quality work. In fact, one can argue that the opposite is true, as depressed individuals are more likely to employ information strategies that are “more active or systematic, [and] detailed” compared to the “relatively passive or nonsystematic” strategies of people with positive affective states (Sinclair & Mark, 1995, as cited in González-Ibáñez & Shah, 2016, p. 2). But the question that concerns us here is how information retrieval systems can enhance the cognitive processes of users who are predisposed to negative affective states based on their past experience in the lifeworld.

The literature on research tasks evoking negative affective states is unequivocal. This phenomenon has even been experimentally measured in healthy control groups, defined by Douglas et al. (2012) as individuals without the presence of a mental illness (p. 279). Significantly, we can understand this to mean that information use predicated on technological systems can adversely affect individuals. Like Kushlev et al.’s findings mentioned at the beginning of this paper, information practices in the digital age can potentially retard social and emotional growth. Such an outcome, however, is ultimately determined by wider systemic factors arising from users’ social and economic environments. Still, the more complex the task, the more difficult it is for users to deduce task inputs, search processes, and search outcomes, which impacts negatively on users’ emotional states (Poddar & Ruthven, 2010, pp. 42-43).

Thus far, only the beginning stages of research have been described and implicated in producing negative feelings among users. But many post-secondary students have an aversion to the writing process as well. This process relies on variegated skills such as elaboration of sources, paper organization, critical evaluation, and synthesis of arguments. Zhou (2013) argues that negative emotions interrupt the regulation and application of these skills. I contend, however, that the presence of negative affective states (even in regard to a psychological disorder like major depression), does not necessarily mean that the user so affected is less willingly engaged in the research process, or that their problem-solving behavior is less identifiable. Judging by the research of Cohen et al., conditions like depression – and negative mood states in general – will only slow the cognitive processing of a user while they are engaged in an information research task. Cognitive insight does not necessarily disappear. The question becomes: can this cognitive slowing be allayed by the right kind of information system design? As we already know, the main objective of an information retrieval system is to provide users with relevant information that can be evaluated quickly and efficiently for a multitude of tasks. Today, in light of the affective research, relevance is no longer a simple measurement of correlation between document representation and user query. In contemporary system design, relevance is a user-centered concept that is concerned with providing a holistic experience where users feel that their retrieved information is valuable and beneficial. I now turn to the prospect of user-centered design based on an implicit understanding of negative affective states.

Implications

Efforts to temper the negative feelings of users began in the field of Human-Computer Interaction through the use of affective language displays. For example, apologetic on-screen display (OSD) messages were programmed into some information retrieval systems to empathize with users in various contexts of the search process. Park et al. (2016) believe that apologetic interaction can help moderate negative emotions and decrease user frustration, as well as increase users’ feelings of system trustworthiness, aesthetic, and usability. Most information retrieval systems, however, are neutral in that they only return status reports of a user’s search. For example, when a search query fails to retrieve any dynamic or static content, many systems will simply display a “No results found” page. Ultimately, Park et al. found that apologetic OSD messages markedly improved users’ levels of frustration with the system (p. 736). Improvements were also found in the perceptual categories of usability and aestheticism. Participants felt that an apologetic system was more usable than a neutral system (p. 737). Additionally, and more surprisingly, participants also felt that the apologetic system was more aesthetically appealing (p. 736). Although this last finding may have more to do with the interface design of the testing systems than the actual presence of affective messages. In any case, it appears that designing systems with emotionally-sensitive messages does have the potential to moderate the information behavior of users. But the authors’ state that “little is known about the impact such affective messages have on users’ affective states and their perceptions of the system” (p. 733). Moreover, it is unknown whether or not these messages have any long-term positive effects on users’ affective character.

Dynamic human-system interaction is an area suitable for artificial intelligence scientists as well as psychologists. Affective messages must be displayed at the right time and in the right context for specific scenarios. For instance, users employ different strategies and keywords when searching for information, so the system has to be dynamic and adaptive to different styles of searching. Furthermore, the system must understand its users in order to provide proper affective messages. This matter increases in complexity when considering the relation between relevance judgments and affective responses. Barral et al. (2016) mention that “after deciding that a text is relevant, one might start reading the relevant text item, which in turn might elicit several emotional responses” (p. 505). This interplay of the cognitive and the emotional is sometimes unpredictable, as it can be difficult to determine in advance the reactions that different individuals will have when confronted with certain pieces of information. For example, a news article editorializing on a national tragedy may bring one individual to tears, while another individual may respond by disappointingly shaking their head. In other words, individuals have unique psychological and personality traits that cause them to react in different ways.

Several studies have attempted to measure physiological responses to information in order to determine users’ felt emotions (e.g. Gwizdka, 2014; and González-Ibáñez & Shah, 2016). Affective recommender systems have been posited (Tkalčić et al., 2011), whereby metadata derived from physiological tests can be used by the system for affective messaging and suggestions for further reading. Barral et al. (2016) annotated textual content with electrodermal data to measure user relevance and emotional response. These types of studies are essential for designing information retrieval systems that can mitigate negative affective responses and redirect users, encouraging them to interact with systems in a more cognitively-enriching way. The question remains, however, if it is ethical to allow machine intelligence the ability to manipulate user perception of information.

Conclusion

The phenomenal lifeworld situates all actions that human beings take in their social and cultural environments. This paper investigated the subjective experience of users with information retrieval systems. The act of gathering and evaluating information has been shown to be rooted biologically, and the way users create meaning out of information retrieved depends on their past experience, as well as their sedimented knowledge and emotions. With the prevalence of digital technologies and networked retrieval systems, the process of information seeking, use, and sharing has become much more insular and isolated in the digital age. The negative emotions that follow complex research tasks are often compounded by this isolation. Because information retrieval systems are increasingly used for everyday information practices, it is important to consider the social integration of these systems. Social phenomenology offers an alternative perspective to approach issues that have so far been discussed in terms of “information behavior” (Savolainen, 2008, p. 202). The intersubjectivity of users and system designers is often obscured through the mediation of systems, but system designers need to be aware of how users approach the information search process. This approach is not performed in a vacuum, as users will always bring their subjective experiences with them to the user interface. Subjective experience is the province of phenomenology, and I think, therefore, that it is a meaningful framework for the information sciences.

References

Barral, O., Kosunen, I., Ruotsalo, T., Spapé, M. M., Eugster, M. J., Ravaja, N., & Kaski, S (2016, November 15). Extracting Relevance and Affect Information from Physiological Text Annotation. User Modeling and User-Adapted Interaction, 26(5), 493-520.

Cohen, A. O., Dellarco, D. V., Breiner, K., Helion, C., Heller, A. S., Rahdar, A., & Pedersen, G (2016, March 1). The Impact of Emotional States on Cognitive Control Circuitry and Function. Journal of Cognitive Neuroscience, 28(3), 446-459.

Douglas, K. M., Porter, R. J., Knight, R. G., & Alsop, B. (2012) The Dynamics of Word Retrieval in Major Depression. Australian & New Zealand Journal of Psychiatry, 47(3), 276-283.

González-Ibáñez, R., & Shah, C. (2016, October 14). Using Affective Signals as Implicit Indicators of Information Relevance and Information Processing Strategies. ASIST, 1-10. Gwizdka, J. (2014, August 26). Characterizing Relevance with Eye-tracking Measures. IIiX ‘2014, 58-67.

Kracker, J. (2002). Research Anxiety and Students’ Perceptions of Research: An Experiment. Part I. Effect of Teaching Kuhlthau’s ISP Model. Journal of the American Society for Information Science and Technology, 53(4), 282-294.

Kracker, J., & Wang, P. (2002, January 22). Research Anxiety and Students’ Perceptions of Research: An Experiment. Part II. Content Analysis of Their Writings on Two Experiences. Journal of the American Society for Information Science and Technology, 53(4), 295-307.

Kuhlthau, C. C. (1991). Inside the Search Process: Information Seeking from the User’s Perspective. Journal of the American Society for Information Science, 42(5), 361-371.

Kushlev, K., Proulx, J., & Dunn, E. W. (2017). Digitally Connected, Socially Disconnected: The Effects of Relying on Technology Rather than other People. Computers in Human Behavior, 76, 68-74.

Nahl, D. (2007, September 7). Social–Biological Information Technology: An Integrated Conceptual Framework. Journal of the American Society for Information Science and Technology, 58(13), 2021-2046.

Park, S. J., MacDonald, C. M., & Khoo, M. (2012, June 11). Do You Care if a Computer Says Sorry? User Experience Design through Affective Messages. DIS 2012.

Poddar, A., & Ruthven, I. (2010, August 18). The Emotional Impact of Search Tasks. Proc. of the 3rd IIiX 2010, 35-44.

Savolainen, R. (2008). Everyday Information Practices: A Social Phenomenological Perspective. Lanham, MD: Scarecrow Press.

Savolainen, R. (2015). The Interplay of Affective and Cognitive Factors in Information Seeking and Use: Comparing Kuhlthau’s and Nahl’s Models. Journal of Documentation71(1), 175-197.

Tkalčić, M., Košir, A., Tasić, J. (2011). Affective Recommender Systems: The Role of Emotions in Recommender Systems. Proceedings The RecSys 2011 Workshop on Human Decision Making in Recommender Systems, 9–13.

Zhou, M. (2013, April 24). ‘‘I am Really Good at It’’ or ‘‘I am Just Feeling Lucky’’: The Effects of Emotions on Information Problem-solving. Association for Educational Communications and Technology, 505-520.

Advertisements

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

w

Connecting to %s