Elsevier

Consciousness and Cognition

Volume 12, Issue 3, September 2003, Pages 452-484
Consciousness and Cognition

Task unrelated thought whilst encoding information

https://doi.org/10.1016/S1053-8100(03)00018-7Get rights and content

Abstract

Task unrelated thought (TUT) refers to thought directed away from the current situation, for example a daydream. Three experiments were conducted on healthy participants, with two broad aims. First, to contrast distributed and encapsulated views of cognition by comparing the encoding of categorical and random lists of words (Experiments One and Two). Second, to examine the consequences of experiencing TUT during study on the subsequent retrieval of information (Experiments One, Two, and Three). Experiments One and Two demonstrated lower levels of TUT and higher levels of word-fragment completion whilst encoding categorical relative to random stimuli, supporting the role of a distributed resource in the maintenance of TUT. In addition the results of all three experiments suggested that experiencing TUT during study had a measurable effect on subsequent retrieval. TUT was associated with increased frequency of false alarms at retrieval (Experiment One). In the subsequent experiments TUT was associated with no advantage to retrieval based on recollection, by manipulating instructions at encoding (Experiment Two), and/or at retrieval (Experiment Three). The implications of the results of all three experiments are discussed in terms of recent accounts of memory retrieval and conscious awareness.

Introduction

When engaged on a task, one’s attention can be directed to information not readily observable in the current environment, and instead we are engaged in the processing of internally generated information such as memories. The attentional shift that accompanies the processing of internally generated information may represent a situation in which the content of conscious awareness becomes, to some extent, decoupled from the processing of ‘external’ perceptual information (Kanwisher, 2001; Merickle, Smilek, & Eastwood, 2001). Kanwisher (2001) suggests that the stimulus features that are associated with the contents of awareness fall into one of the following three categories: (i) low level stimulus features such as stimulus detection, (ii) mid level features such as stimulus shape or (iii) the higher level meaning of the stimuli (Kanwisher, 2001). It is likely that these features are, at least to some extent, served by different neural systems, and the understanding of the interaction of these different codes of neural information may be important in our understanding of conscious awareness (see for example the consciousness as fame metaphor, Dennet, 2001).

This paper describes three experiments that investigated the consequences of processing internally generated information whilst ostensibly engaged in a task to encode verbal information. In these experiments a shift in attention was measured through the use of a thought probe which terminates the current block of the task, as has been described elsewhere (Smallwood, Obonsawin, Reid, & Heim, in press-A; Smallwood, Obonsawin, & Reid, in press-B). Using this paradigm, thoughts are recorded verbatim and classified in terms of whether they reflect attention to matters unrelated to the task in hand or the current situation task unrelated thought (TUT, see Smallwood et al., in press-B for detailed criteria on making these judgements).1 One focus for this article is the contrast between encapsulated and distributed accounts of cognition, in terms of their predictions for the distribution of TUT. The second is the analysis of the empirical consequences of experiencing TUT on memory retrieval, particularly focusing on the contrast between the influences of recollection and familiarity on retrieval from memory (Jacoby, 1998; Jennings & Jacoby, 1993).

Maintaining ones attention on the current task has been historically ascribed to the resources directed to task completion (e.g., Giambra, 1995). For example, it has been suggested that TUT was the product of a limited all-purpose cognitive resource (Antrobus, 1968) and similar models have been suggested to explain the ‘unitary’ nature of consciousness (e.g., Mandler, 1975; Posner & Kline, 1973). Broadly, encapsulated models of processing propose that the amount of TUT experienced during any given task is inversely proportional to the amount of resources deployed towards the task. In this context, ‘Effort’ may be conceptualised in terms of bits of information processed per second (Giambra, 1995).

Empirical research into TUT broadly supports the prediction of an encapsulated resource (Antrobus, 1968; Giambra, 1995). There is, for example, a robust and replicable influence of presentation rate of stimuli on TUT production during vigilance (e.g., Antrobus, 1968; Giambra, 1995). In addition, increasing the size of a digit string to be shadowed decreases the report of TUT whilst shadowing numerical information (Teasdale, Lloyd, Proctor, & Baddeley, 1993). Viewed from the perspective of an encapsulated system, qualitative differences in the contents of awareness stem from quantitative differences in the deployment of attentional and cognitive resources. These changes can be ascribed to changes in low-level stimulus features (Kanwisher, 2001).

The notion of schema arose in cognitive psychology out of a need to recognise that higher-level structures were necessary to describe the complex relations implicit within our knowledge base (Minsky, 1975). Schematic processing has been demonstrated to play an important role in a wide range of processes, such as memory (Brewer & Treyens, 1981), the maintenance of social norms via the use of scripts (Shanck & Abelson, 1977) and text comprehension (Westby, 1999). A schematic approach to cognition provides a mechanism whereby the global state of the system, the context of information processing, can facilitate the processing of local information.

Accounts of cognition which are consistent with the role of global states in information processing (such as schemata) are those models that can be considered ‘distributed’ (e.g., Faulconnier & Turner, 1998; Kennephol, 1999; Rumelhart & Mclleland, 1986). A well-established example of a distributed account of cognition is parallel distributed processing (PDP, Rumelhart et al., 1986). The main features of a PDP network are generally agreed upon (Kennephol, 1999; Smolensky, 1999). Kennephol (1999) identifies three broad characteristics of PDP networks. In PDP networks: (i) information processing occurs in parallel, (ii) representations are not associated with a single entity, rather they are distributed across the network as a whole, and (iii) concepts are represented by activation patterns across the system as a whole therefore avoiding the necessity of a central executive (Kennephol, 1999).

One area where distributed models of cognition are particularly relevant is in the generation and maintenance of categories (see Pinker, 1997). The distributed approach to cognition is consistent with this prototypical view of category membership (see Barsalou, Huttenlocher, & Lamberts, 1998; Mandeblit & Zachar, 1998, pp. 243–247; Rumelhart et al., 1986, pp. 183–192; see also Wittgenstein, 1953) as both the prototypical view and the distributed approach share many assumptions regarding categorical perception. For example, both the connectionist view and the prototype view propose that category membership depends upon frequent experience of category members, a position supported by empirical evidence (Barsalou et al., 1998).

Recent evidence has examined whether contextual2 features can affect the experience of TUT. One contextual feature of relevance is the goal state of the individual: their current concerns.3 Ecological evidence suggests that as individuals go about their day-to-day lives they are often pre-occupied with their current concerns (Klinger, Barta, & Maxeiner, 1980). Moreover, overlap between questionnaires measures of daydreaming and coping strategy, implies that daydreaming may function to re-enforce coping behaviours (Greenwald and Harder, 1995, Greenwald and Harder, 1997). Evidence from a laboratory situation is consistent with these findings (Klinger, 1978). Participants were engaged in a dichotic listening task with two streams of verbal information, one of which was relevant to the individual’s current concerns, and one that was not. The results suggested that participants spent more time listening to the tape reflecting their current concerns and reported more thoughts relevant to those issues (Klinger, 1978). It seems plausible, therefore, that the experience of TUT may “serve as a continual reminder of the rest of the persons agenda” (Klinger, 1999, p. 439). In one important aspect, therefore, TUT is clearly affected by the internal context of the individual.

It has been suggested that TUT is a consequence of the individual’s goals and motivation and is, therefore, likely to be associated with functional consequences, such as facilitating problem solving by the conceptual manipulation of semantic information (Baddeley, 1999; Binder et al., 1999). Binder and colleagues investigated the role conceptual semantic processing plays in the experience of TUT. During vigilance, TUT was reported less frequently than during an eyes-shut control condition. The results of a subsequent fMRI experiment suggested that a network of left-hemisphere poly-modal cortical regions, including the pre-frontal cortex, showed lower levels of activation during vigilance than either during the resting state or the semantic processing task (Binder et al., 1999). These findings suggest that similar “conceptual” activation is involved in the on-going, task unrelated, conscious activity during the resting state as is involved during the semantic processing task. Binder and colleagues suggest that such a capacity has a functional basis: “By storing and manipulating internal information [of a conceptual nature] we organise what could not be organised during stimulus presentation, solve problems that require computation over long periods of time, and create effective plans governing behaviour in the future. These capabilities have surely made no small contribution to human survival and the invention of technology” (Binder et al., 1999, p. 85).

Smallwood et al. (2003) attempted to clarify the role of conceptual processing by comparing the experience of TUT whilst processing information organised either alphabetically or in meaningful semantic categories. These experiments allowed a contrast between the predictions of a distributed resource with an encapsulated model of cognition in terms of their predictions regarding the frequency of TUT. In an encapsulated model, categorical processing is considered less effortful than non-categorical processing because categories/schemata represent a quality of information which are “highly organized, having a clear thematic structure” (Ellis & Ashbrook, 1986, p. 31). Based on an encapsulated view of the mind, therefore, categorical processing should yield higher frequencies of TUT. In contrast, a distributed model emphasises the context of information processing in mediating task focus. In the case of processing information organised into categories, the likelihood that the internal context of information processing corresponds to the external environment would be higher, making the experience of TUT less likely (Smallwood et al., 2003).

During both fluency and memory tasks, participants performed at a superior level and reported a lower frequency of TUT in the categorical tasks (Smallwood et al., 2003). This pattern of results is consistent with the role of a distributed resource in the generation of TUT. A possible problem with the experiments, however, was that the stimulus set used in each condition, categorical or alphabetical, was different prohibiting robust conclusions being drawn on the role of stimulus organisation (Smallwood et al., 2003).

The first aim of this paper was, therefore, to replicate and extend our knowledge of the relationship between categorical processing and TUT. In particular, this paper examined whether the categorical suppression of TUT was observed (i) when differences in the stimuli set are controlled for (Experiment One) and (ii) to examine the extent to which the suppression of TUT is concerned with processing the stimulus set per se, or is better understood as an affordance4 between task and stimulus materials (Experiment Two).

If TUT reflects a shift in conscious awareness towards the processing of an internal source of information at the expense of exogenous attention, one would expect that this shift in information processing would co-occur with a change in external information processing. Moreover, given that the dependent measure in this series of experiments is subjectively reported information, and that the base rate for occurrence of these phenomena whilst encoding information is low (Smallwood et al., 2003, Smallwood et al., in press-B; Teasdale et al., 1993) it is clearly important to validate the subjective information reported by the individuals during task completion by reference to measurable aspects of task processing (Baars, 1988).5 Recently evidence has demonstrated subtle, measurable detriments in task performance that accompany the experience of TUT (Teasdale et al., 1995, Study Four; Smallwood et al., in press-B). For example, Teasdale et al. (1995) suggested that the experience of TUT during a random number generation task was reflected in a decrease in the random nature of the numbers generated. More recently in the context of memory tasks, a positive correlation was observed between the report of TUT and the report of false alarms during recall (Smallwood et al., in press-B).

As TUT has been shown to be positively associated with higher levels of false alarms at the point of recall, it seems plausible that TUT during study may effect the encoding of information. It is generally accepted in the literature that retrieval can occur through two distinct routes. These two routes are described as recollection vs. familiarity (Jacoby, 1998; Jennings & Jacoby, 1993) or explicit vs. implicit (Szymanski & MacCleod, 1996). Broadly, the influences of recollection/explicit recall can be understood as the “conscious retrieval of an event” (Szymanski & MacCleod, 1996, p. 165) whilst familiarity/implicit influences occur “when a previously encountered stimulus affects behaviour without conscious knowledge” (Szymanski & MacCleod, 1996, p. 165).6 One important influence on recollection/familiarity is the nature of attention to the stimulus at the point of study. Both divided attention (Jacoby, 1998; Jennings & Jacoby, 1993; Parkin, Reid, & Russo, 1990) and directing attention to superficial stimulus features (Szymanski & MacCleod, 1996) seem to impair stimulus recollection, whilst leaving the effects of stimulus familiarity unaffected.7 As TUT reflects the explicit description by the participant that the contents of their attention departed from the current situation, it seems plausible that experiencing TUT during study may effect one or other component of subsequent retrieval: recollection or stimulus familiarity.

The second broad aim of these experiments is, therefore, to examine the consequences of experiencing TUT during study on the subsequent retrieval of information. In Experiment One we compared those individuals who report one or more TUT during study with those individuals who reported none, in an attempt to replicate the positive association between false alarms and TUT reported elsewhere (Smallwood et al., in press-A). Experiments Two and Three examine whether the effects of TUT on study of the subsequent retrieval of information can be understood in terms of the recollection/familiarity distinction. This was achieved by manipulating instructions at encoding (Experiment Two) and by manipulating instructions at retrieval (Experiment Three) using the process dissociation procedures developed by Jacoby and colleagues (Jacoby, 1998; Jacoby, Toth, & Yonelinas, 1993).

Section snippets

Experiment One

The aims of Experiment One are as follows. First to demonstrate that the categorical suppression of TUT reported in Smallwood et al. (2003) was not an artefact of the stimuli set used. We examined the frequency of TUT during the encoding of two sets of stimuli as a within-participant factor: one set of stimuli was presented in blocks organised categorically, the other set organised in a random fashion. In line with previous research, it was expected that TUT will be lower during the categorical

Experiment Two

Several issues remain unclear regarding the findings of Experiment One. First, a consistent picture is developing that TUT is inhibited when processing a categorical stimulus organisation (Smallwood et al., 2003). These demonstrations use demanding tasks (either fluency or memory tasks) and task process has been acknowledged to be an important component of maintaining task focus (Teasdale et al., 1993). On the basis of the results of Experiment One alone it is unclear whether the decrease in

Experiment Three

In the literature on retrieval from memory, there is some debate about whether it is possible to separate the recollection/familiarity influences by task instructions (see Jacoby, 1998). Advocates of this position (Jacoby, 1998; Jacoby et al., 1993) suggest that, irrespective of instructions, all tasks yield retrieval that is based on a combination of recollection and familiarity. To reconcile this view, Jacoby and colleagues suggest it is possible to assess the different components of

Methodological limitations

Before dealing with the implications of these findings it is worth dealing with several possible problems with the interpretation of the results. A key problem with the data presented throughout this paper is the low base rate in the production of TUT. Whilst the frequency of TUT is higher than that reported by Smallwood et al. (2003) as a result of a higher number of blocks each participant was engaged in, the categorical conditions, in particular, yield lower base rates than described

TUT and memory retrieval

It is clear that TUT at study is associated with differences in the subsequent retrieval of information. On the one hand, TUT is generally experienced in those situations in which encoding is impaired (such as when encoding non-categorical information, Experiments One and Two; Smallwood et al., 2003, Experiment Three). In a similar fashion, when TUT is experienced it is associated with retrieval from memory in which recollection is not statistically different from retrieval based on

Acknowledgements

The authors thank Larry Jacoby for his advice on the design of the third experiment, and the comments of two anonymous reviewers on an earlier version of the manuscript. Thanks to Gordon McAlpine for writing the computer programme used in these experiments. Finally, we wish to thank the participants who took part in these experiments.

References (61)

  • W.F. Battig et al.

    Category norms for verbal items in 56 categories

    Monograph of the Journal of Experimental Psychology

    (1969)
  • J.R. Binder et al.

    Conceptual processing during the conscious resting state: A functional MRI study

    Journal of Cognitive Neurosciences

    (1999)
  • F. Crick

    The astonishing hypothesis: The scientific search for the soul

    (1994)
  • J.B. Davies

    The myth of addiction

    (1992)
  • D.C. Dennet

    Are we explaining consciousness yet?

    Cognition

    (2001)
  • Ellis, H. C., & Ashbrook, P. W. (1986). Resource allocation model of the effects of depressed moods states on memory....
  • G. Faulconnier et al.

    Conceptual integration networks

    Cognitive Science

    (1998)
  • D.F. Greenwald et al.

    Sustaining fantasies, daydreams and psychopathology

    Journal of Clinical Psychology

    (1995)
  • D.F. Greenwald et al.

    Fantasies, coping behaviour and psychopathology

    Journal of Clinical Psychology

    (1997)
  • A. Grodsky et al.

    Task unrelated images and thoughts whilst reading

  • S. Hartlage et al.

    Automatic and effortful processing in depression

    Psychological Bulletin

    (1993)
  • L.L. Jacoby

    Invariance in automatic influences of memory: Towards a users guide for the process-dissociation procedure

    Journal of Experimental Psychology: Learning Memory and Cognition

    (1998)
  • L.L. Jacoby et al.

    Separating conscious and unconscious influences of memory: Measuring recollection

    Journal of experimental Psychology: General

    (1993)
  • J.M. Jennings et al.

    Automatic versus intentional uses of memory: Aging, attention, and control

    Psychology and Aging

    (1993)
  • S. Kennephol

    Towards a cultural neuropsychology: An alternative view and a preliminary model

    Brain and Cognition

    (1999)
  • E.C. Klinger

    Modes of normal conscious flow

  • E.C. Klinger

    Thought flow: Properties and mechanisms underlying shifts in content

  • E. Klinger et al.

    Motivational correlates of thought, content, frequency and commitment

    Journal of Personality and Social Psychology

    (1980)
  • A. Mack et al.

    Inattentional blindness

    (1998)
  • C.M. MacLeod

    Half a century of research on the stroop effect: An integrative review

    Psychological Bulletin

    (1991)
  • Cited by (206)

    View all citing articles on Scopus
    View full text