Temporal Immediacy

Abstract

Background:

Health promotion and chronic disease management both require behavior change, but people find it hard to change behavior despite having good intentions. The problem arises because patients’ narratives about experiences and intentions are filtered through memory and language. These narratives inaccurately reflect intuitive decision-making or actual behaviors.

Objectives:

We propose a principle—temporal immediacy—as a moderator variable that explains which of two mental systems (narrative or intuitive) will be activated in any given situation. We reviewed multiple scientific areas to test temporal immediacy as an explanation for findings.

Methods:

In an iterative process, we used evidence from philosophy, cognitive neuroscience, behavioral economics, symptom science, and ecological momentary assessment to develop our theoretical perspective. These perspectives each suggest two cognitive systems that differ in their level of temporal immediacy: an intuitive system that produces behavior in response to everyday states and a narrative system that interprets and explains these experiences after the fact.

Findings:

Writers from Plato onward describe two competing influences on behavior—often with moral overtones. People tend to identify with the language-based narrative system and blame unhelpful results on the less accessible intuitive system, but neither is completely rational, and the intuitive system has strengths based on speed and serial processing. The systems differ based on temporal immediacy—the description of an experience as either “now” or “usually”—with the intuitive system generating behaviors automatically in real time and the narrative system producing beliefs about the past or future.

Discussion:

The principle of temporal immediacy is a tool to integrate nursing science with other disciplinary traditions and to improve research and practice. Interventions should build on each system’s strengths, rather than treating the intuitive system as a barrier for the narrative system to overcome. Nursing researchers need to study the roles and effects of both systems.

Over 50% of deaths in the United States are attributable to chronic diseases (Jiaquan, Shery, Kenneth, & Brigham, 2016), many of which are directly linked to health behaviors such as diet, exercise, and smoking (World Health Organization, 2016a, 2016b). Health behaviors have accordingly been an important focus of nursing theories and interventions (Glanz, Rimer, & Viswanath, 2015). Patient self-management has been identified as one key to achieving the triple aim of better quality, lower cost, and improved satisfaction in healthcare (Berenson, Devers, & Burton, 2011), with some evidence that patients who actively manage their health have lower total costs of care for chronic diseases (Hibbard, Greene, & Overton, 2013).

Despite research-supported models for predicting health behaviors and interventions for modifying them, health behaviors seem resistant to change (Dunkley et al., 2014; Ling, Robbins, Wen, & Zhang, 2017; Nieuwlatt et al., 2014; Portnoy, Scott-Sheldon, Johnson, & Carey, 2008). The paradox that people fail to do what is good for them has been noted throughout history, from Aesop’s fables to recent behavioral economics studies (Kahneman & Tversky, 2000). This is a vexing problem: Nurses want to help people live healthier lives, yet people often seem resistant to efforts to make them more aware, empowered, and engaged. Similarly, top health psychology researchers expressed frustration that, despite their field’s initial promise, interventions seem to have reached a ceiling efficacy level (Rothman, 2016). The intransigence of health behavior problems must be one starting point for any useful theory to address them.

Any valid theory of health behavior must also address the fact that patients themselves experience discrepancies between their stated intentions and actual behaviors. This observation also dates from antiquity, as in the words of St. Paul: “The good that I would I do not; but the evil which I would not, that I do” (Romans 7:19). Contemporary health behavior models confront this problem in the form of unexplained variance: For instance, despite the predictive ability and wide dissemination of Ajzen’s (2011) theory of planned behavior (TPB), even the author says that his model has “natural limitations” because of discrepancies between intentions and behaviors. In fact, Ogden (2003) found the TPB to be predictive only when people said they did not intend to change; when they did intend to engage in healthy behavior, the TPB was no better than chance. Similarly, Herzog and Blagg (2007) found that questions about intentions are less useful than simply asking people to predict their future behaviors. A surprising number of people have the insight, “I do intend to do this, but I also can tell you right now that I won’t.”

Purpose, Aims, and Process for Theory Development

A major reason for the observed discrepancy between people’s intentions and behaviors lies in the difference between people’s intuitive responses to present experiences and their later narratives about them. People commonly attribute discrepancies between intention and behavior to a “lack of willpower” (Heider, 1958), but this assumption ignores other factors that behavioral economics and cognitive neuroscience have begun to uncover. Below, we present evidence that intentions and behaviors arise from two distinct modes of thinking and from two distinct neuropsychological processes. Most theories to explain exercise, diet, or treatment adherence were formulated and tested using questionnaire measures of intentions or behaviors. These rely on an abstract mode of thinking in which people make judgments based on their recollections or expectations. However, everyday experience is more immediate, less reflective, and mostly nonverbal. Unfortunately, people’s summary judgments correlate poorly with data on the same constructs collected in real time using technology-supported measures (Shiffman, Stone, & Hufford, 2008).

We used an iterative process of theory synthesis (Walker & Avant, 2011) to integrate converging findings across disciplines that explain behavior in terms of two mental systems (Table 1). Starting from observations and challenges that three of the authors encountered in our own research, we examined multiple two-system models (Evans, 2008; Kahneman, 2011; Pacherie, 2006) in an effort to understand the intention–behavior gap. We developed a novel contribution to the two-system theory, the idea that people’s two minds can be operationally distinguished based on temporal immediacy and the quality of experience as now versus past or future. We then tested whether our tentative conclusions fit with recent observations from (a) philosophy of consciousness, (b) cognitive neuroscience, (c) behavioral economics, (d) nursing symptom science, and (e) ecological momentary assessment studies. At each stage, we considered how our view of humans’ two mental systems might need refinement to accommodate discrepant or converging findings. Late in the process, we found that our evolving view of the nonconscious system that generates behavior—what we call the intuitive system—seemed to fit with recent ideas on mindfulness as a way to access intuitive wisdom (Paulson, Davidson, Jha, & Kabat-Zinn, 2013), and we invited a mindfulness expert (the fourth author) to comment. Once we were satisfied that our model provided a reasonably accurate explanation for intention–behavior gaps and was consistent with findings in each of these generally unconnected domains of knowledge, we explored clinical and research implications of our proposed model. At each stage of our synthesis, we found that people’s immediate present experiences—what we describe as the intuitive system—are distinct from language-based labels or summaries of experience—what we call the narrative system—to encompass its imagining and remembering roles. We suggest that intuitive system predictors of behavior differ from those in well-known models based on self-reports from the narrative system and that this difference can explain the intention–behavior gap.

TABLE 1.

Process of Theory Synthesis

Step	Sources of information and process of synthesis	Rationale and conclusions
1	Nursing symptom science: The first, fifth, and sixth authors have each led studies of symptom science and previously collaborated on a method for understanding reciprocal relationships among bio behavioral phenomena such as symptoms or health behaviors (Corwin, Meek, Cook, Lowe, & Sousa, 2012). Our “shape shifters” model suggested that such phenomena are both psychological and physiological, with neither aspect being primary, and that a given variable is either a “phenomenon” (influenced by other variables) or a “determinant” (cause of other variables) depending on research design.	• Osn the basis of this starting point, we viewed health behavior as wholly determined by neither attitudes and beliefs as in cognitive—behavioral models nor physiology as in neuroscience views of thought as an epiphenomenon. Rather, we were looking for a model that could successfully explain empirically observed health behaviors at both levels.
2	Ecological momentary assessment: Another important set of inputs came from the same three authors’ programs of research on patients’ daily experiences. In studies of symptoms and behaviors, we had each independently observed that people’s experiences or behaviors in the moment are often quite different from what they have said they expected or from what they remember after the fact.	• We found problems of this type both in studies of symptom experiences and in studies of health behaviors, and our own findings were replicated in the literature. The observed gap between immediate and recalled experience was the phenomenon we set out to explain.
3	Philosophy of consciousness: When trying to explain and predict human behavior, we believe it is naive to ignore the thousands of years of expertise developed by the world’s great religions and philosophical traditions. These schools of thought provide many relevant observations about humans’ experiences across cultures and over time, and they continue to subtly shape modern scientists’ thinking.	• Like others, we found that “all of Western thought is a series of footnotes to Plato” (Goldstein, 2014). We were interested to see how Plato’s two-horse theory shaped later Western dualistic thinking and added moral overtones to modern two-system theories.
4	Behavioral economics: Consideration of two-system theories led us back to Kahneman’s (2011) prospect theory, which we were familiar with but had not previously considered in detail. We found much to commend this theory but also some of the original problems from Plato’s two-horse theory. Kahneman also proposed a second, unrelated distinction between experience and memory. This idea was the turning point in developing our theory, and after formulating a model, we returned to each earlier step to test our ideas.	• Kahneman’s theory clearly helps to explain gaps between intentions and behavior, but (a) it fails to explain why rationality is selectively active (it requires a homunculus) and (b) it maintains the morally superior character of narrative thought. The experience-memory distinction instead suggests two systems that act at different times.
5	Cognitive neuroscience: Because we have a biobehavioral understanding of human nature, we next looked for physiological processes that might mirror Kahneman’s two systems. We found better support for Kahneman’s “experiencing versus remembering self” dichotomy than for his “rational versus emotional” distinction between systems. The best physiological evidence suggests no causal role for rationality in producing many behaviors.	• At this point, our idea of temporal immediacy crystallized as a way to resolve problems with Kahneman’s two-system view and to explain intention-behavior gaps. Findings from each of the five areas that we had considered so far were consistent with our theory.
6	Mindfulness: The idea of automatic, intuitive system processes also brought to mind the idea of mindfulness, and we invited an additional coauthor with expertise in mindfulness research to join us. This led to important elaboration in multiple areas of our thinking.	• The fourth author pointed out further examples of positive outcomes from intuitive system processes and other examples of temporal immediacy.
7	Research and practice implications: Finally, we returned full circle to the research methodologies that first let us to observe inconsistencies between immediate and recalled experiences or between intentions and behaviors, to see how they might be augmented.	• The second and third authors are experts in monitoring technologies and statistical methods, who helped us to explore the implications of our proposed model.

Two Horses: The Classical View of Intention–Behavior Discrepancies

The philosopher Plato proposed a metaphor in which “the charioteer of the human soul drives a pair [of winged horses]. One of the horses is noble and of noble breed, but the other quite the opposite in breed and character.… The horse of evil nature weighs the chariot down, making it heavy and pulling toward the earth the charioteer whose horse is not well trained…[while the noble horse] needs no whip, but is guided only by the word of command and by reason” (Phaedrus, pp. 246–254). Traditionally, the noble horse represents reason, and the evil-natured horse represents emotions or physiological drives. This concept is the basis for so much later thought that most people in Western societies assume a tension between reason and emotion—even if we have never heard the actual metaphor of the horses before (Goldstein, 2014).

In Plato’s model, people’s actions are pulled in two directions by competing rationality and emotions, and stronger-willed people make better choices. Christian thinkers like St. Augustine of Hippo (2003) built on Plato’s ideas as a metaphor for struggle between good and evil in human nature. Martin Luther (1525/2012) espoused a similar dualism but attributed unhealthy impulses to a force literally outside himself (i.e., the devil), while crediting all positive impulses to God. A modern variation explains intention–behavior gaps based on the biology of willpower: In “marshmallow test” experiments, toddlers who resisted eating a marshmallow had more brain activity in areas linked to self-control—the prefrontal cortex (PFC) and ventral striatum—and greater success in later life (Casey et al., 2011). Cognitive–behavioral methods are also built on Plato’s propositions, using behaviorist strategies to tame the wild horse (Skinner, 1990) and rational arguments win the noble horse’s cooperation (Beck, 1995). The common theme in all these models is that they see humans as torn between base impulses and noble rationality, in an ongoing struggle to overcome the temptations of daily life.

A Horse of a Different Color: Two-System Theories

An important recent revisioning of Plato’s theory of mind comes from Nobel laureate Daniel Kahneman, whose book Thinking, Fast and Slow (Kahneman, 2011) popularized the idea that people have two separate “minds.” System 1, which we label the intuitive system, is a set of “effortlessly originating impressions and feelings” that has “freewheeling impulses and associations” and “operates automatically and quickly, with little or no effort and no sense of voluntary control” (p. 20). By contrast, System 2, which we label the narrative system, “allocates attention to the effortful mental activities that demand it…[and is] associated with the subjective experience of agency, choice, and concentration.” Kahneman argued that “only the slower narrative system constructs thoughts in an orderly series of steps” (p. 21), whereas the intuitive system is driven by heuristic decision rules and cognitive biases. There are clearly parallels here to Plato’s two horses: two influences on behavior, one rational and the other not, pulling in different directions. However, there are also important differences from Plato’s ideas.

First, although the intuitive system is often described as the less rational driver of behavior, it also has important strengths. Intuitive thinking is quick and effortless and can deliver excellent results even without conscious awareness. For instance, intuitive processes incorporate learning, as seen in people’s ability to drive cars, the swing of golf professionals, or the winning instincts of chess masters (Lisman & Sternberg, 2013). Intuitive thinking also leads people to resist injustice based on a process that is fast and separate from rational arguments (Bocchiaro & Zimbardo, 2017). Although narrative thinking has been identified with reason and is often believed to produce better results, Kahneman (2011) acknowledged that this system also sometimes makes errors—for example, 5% mortality is rated as a worse outcome than 95% survival despite their logical equivalence—and is limited by “laziness” or chronic inactivity.

Second, Kahneman (2011, p. 21) linked the narrative system to awareness, “the conscious, reasoning self that has beliefs, makes choices, and decides what to think about and what to do.” This is quite different from the metaphor of two horses, where the horse of divine nature and the horse of earthly nature were both subject to human conscience or will—the charioteer of Plato’s model. In contrast to older Western thought in which a separate moral “me” decided between the two impulses, Kahneman’s version proposed no charioteer. In this view, the more rational horse is what a person calls “me,” and the irrational horse is mostly outside his or her awareness. Kahneman’s view aligns better with Buddhist ideas in which narrative thoughts are like a dream and people live without full awareness of their experiences (Paulson et al., 2013). Another description of the two systems comes from Sigmund Freud, who called the narrative self “I” (ego) and the intuitive self “it” (id; Fancher, 1973). Yet, both of these modes of thinking are in fact “I,” although most people typically identify themselves only with the narrative system.

Finally, in proposing a “lazy” narrative system, Kahneman (2011) made the important suggestion that the intuitive system is generally more active in daily life, with the narrative system playing a role only at specific times or in specific contexts, such as when making future plans or completing a questionnaire. Kahneman suggested a sort of veto power, in which “System 1 continuously generates suggestions for System 2: impressions, intuitions, intentions, and feelings. If endorsed by System 2, impressions and intuitions turn into beliefs, and impulses turn into voluntary actions” (p. 24). Instead of an active executive, Kahneman characterized the narrative system as an ineffectual parent that distractedly keeps track of the intuitive system and intervenes only when things get out of hand. Although the idea that the narrative system is only selectively active is a major deviation from classical Western thought, we again note that this idea is compatible with Buddhist notions of automaticity (Paulson et al., 2013). It also is consistent with recent research suggesting that the brain’s frontal lobe exercises “executive control” only in the sense of evaluating alternatives, not making decisions (Stanovich, 2009).

Changing Horses: The Theory of Temporal Immediacy

Table 2 summarizes our own two-system theory. Despite the influence of Kahneman’s (2011) work, it has one substantial weakness: There is no clear neuropsychological mechanism for determining when the narrative system will be activated or not (Zimmerman, 2016). In other words, how does the intuitive system know it is in over its head, or when does the narrative system awake from its stupor? Kahneman admitted to falsely anthropomorphizing the systems, which obscures the fact that these are psychological processes and must follow predictable rules. This is a homunculus problem, requiring a separate third mechanism—like Plato’s charioteer—to adjudicate between the two systems in any given situation. Kahneman also clearly prioritized narrative over intuitive processes, fitting comfortably into the long tradition of Western thought that assumes an internal conflict between reason and unreason. Like Plato, Freud, Augustine, and Luther, it appears that Kahneman’s best advice for overcoming the intention–behavior gap is to “try harder.” However, in fact, most successful applications of behavioral economics simply trick the intuitive system into healthier decisions, rather than following Kahneman’s advice to bring logic or reason into play. These problems with Kahneman’s theory make it less useful in practice.

TABLE 2.

Differences Between Intuitive and Narrative Thinking

Intuitive system

Narrative system

Corresponds to Kahneman’s (2011) “System 1” Continuous and multiple (a set of systems) Usually outside conscious awareness, not expressed in language Involvement of subcortical and limbic system areas, including emotional arousal centers Involves habits and first reactions Uses heuristic rules that are usually adaptive but can produce results inconsistent with formal logic Can be trained Uses inductive, bottom-up reasoning to arrive at conclusions Prone to mistakes of omission Can be fooled by contextual cues, first impressions, or isolated features of experience Best at implementing simple rules, approximating, and satisficing based on available alternatives Massively parallel processing—few functional limits to automatic thinking Can produce exceptional results through practice and automatic response sequences Our proposal: also refers to Kahneman’s “experiencing self” and produces actual behavior; focused on the immediate present, with no conscious awareness of the past or future Our proposal: selects behavior automatically based on its expected consequences

Corresponds to Kahneman’s (2011) “System 2” Episodic (Kahneman says “lazy”) Usually involves conscious thought, expressed in language or imagery Involvement of cortical areas, including language, visual interpretation, and social imagination Involves the sense of a continuous self Develops and uses mental scenarios, which can suggest a different course than intuitive processes Can be convinced Uses deductive, top-down reasoning to build support for positions Prone to mistakes of commission Can be fooled by false beliefs, stereotypes, or personal investment in specific results Best at implementing complex algorithms, measuring, and optimizing results Single-focus serial processing—attention is a very limited resource Can produce exceptional results through slow, logical, and sequential thought Our proposal: also refers to Kahneman’s “remembering self” and is active only in planning an action (future) or remembering an action (past); generally has little awareness of the present Our proposal: produces alternative narratives about possible behaviors and their consequences

We now propose a solution to the intention–behavior gap that does not rely on engaging the narrative system in everyday decision-making. We argue that using the narrative system in these contexts is actually impossible because the narrative system works only at a higher level of abstraction. However, our proposed solution also owes a debt to Kahneman (2011), who identified a separate distinction between what he called the “experiencing self” and the “remembering self.” According to Kahneman, “the experiencing self is the one that answers the question ‘does it hurt now?’, while the remembering self is the one that answers the question ‘how was it, on the whole?’” (p. 381). Kahneman was careful to differentiate these two “selves” from his Systems 1 and 2, yet he also stated that “memories are all we get to keep from our experience of living, and the only perspective that we can adopt as we think about our lives is therefore that of the remembering self” (p. 382). This statement seems to align the remembering self with the “I” of the narrative system. In contrast to Kahneman, we propose that the experiencing self actually is the intuitive system and the remembering self is the narrative system. Although this may initially seem to be a trivial semantic distinction, it has important implications. It means that the more “rational” narrative system actually plays no role in generating behavior but only generates commentary about what has happened or what is going to happen. In brief, we argue that the intuitive system deals with the present, that only this system has the immediacy needed to control behavior, and that the narrative system’s only role is to tell stories about the future or the past. Kahneman argued convincingly that focusing on narratives can increase happiness when thinking about a past event like a vacation, but unfortunately, one’s long-term health depends much more strongly on one’s actual behaviors than on the stories one tells oneself about those behaviors. The heavy drinker who thinks “most of my friends also drank this much” may be perfectly content but remains at risk for negative health consequences. Actual behavior matters most in the long term, and we argue that behavior is governed by the intuitive system.

Our own theory is a modification of the two-system theory based on a new concept, which we call temporal immediacy. First and most importantly, we differ from Kahneman in suggesting that the two systems are not in competition for control of behavior; the actual production of behavior is in the domain of the intuitive system alone, and the narrative system is limited to commentary on past behaviors or prediction of future behaviors. This resolves the conflict noted by Zimmerman because each system is active only in its own domain: Plato’s chariot is pulled by a single horse. Second, we argue that the results obtained in any study of behavior will depend on which system was engaged by the measurements used: the narrative system in typical studies using retrospective questionnaires but the intuitive system in studies using direct observation of behavior. Finally, we propose a concept—temporal immediacy—to distinguish between the two systems. Temporal immediacy is an organizing principle that determines which system is likely to be active in a given context. Temporal immediacy is not a characteristic of persons or a typical psychometric construct—one cannot inherit it, develop it, or use it as a way to give the narrative system more control over behavior. Instead, it is a descriptor of the research study, an artifact of design that affects the results. Our overall argument is that paradigms involving less immediacy, such as retrospective questionnaires, will elicit a narrative system response in which people believe their intentions to be logically connected to their behaviors. However, research designs with greater immediacy, such as behavioral monitoring technologies, will elicit primarily an intuitive system response, showing the ground truth of people’s experiences and behaviors.

Hearing Hoofbeats: Evidence Consistent With the Theory

Figure 1 illustrates divisions between the intuitive and narrative systems as well as potential points of connection. After modifying Kahneman’s (2011) theory by adding a temporal differentiation between the two systems, we tested our model by examining relevant evidence from five different domains of knowledge: contemporary philosophy of consciousness, cognitive neuroscience, behavioral economic, nursing symptom science, and ecological momentary assessment. We found convergent evidence from multiple, unrelated areas of study, including an ability to explain Kahneman’s own findings in a new way.

FIGURE 1.

A two-system model based on temporal immediacy. Temporal immediacy, shown by a semipermeable line in the center of the figure, is a contextual feature that determines whether intuitive or narrative system responses are evoked. Although a single stimulus may elicit reactions from both systems, the selective recording of outputs from one system or the other can produce discrepant research results. The lower half of the diagram represents the intuitive system that focuses on the immediate present, including registrations from perceptual systems, automatic responses or instincts that might originate in noncortical brain areas, procedural memory for learned habits or patterns of behavior that do not require conscious thought, and a mechanism for activation of behavior based on its projected consequences that we suggest is exclusively a function of the intuitive system. Potential behaviors come from the intuitive system’s procedural memory and automatic responses but can also include imagined responses generated by the narrative system. The upper half of the diagram shows elements of the narrative system, which is a more conscious, slow, and deliberate mode of thinking that depends on the prefrontal cortex. We suggest that this system contains judgments about experiences that are distinct from the sensory experiences themselves, conscious experiences in which people are able to articulate their own subjective perceptions, declarative memories for stored narratives (“knowing that” as opposed to the intuitive system’s “knowing how”), and narratives that summarize experiences over time. We also propose that attention is part of the narrative system, based on the fact that people have conscious awareness of directing or focusing their attention. The contents of attention might be narratives and judgments or might get closer to actual sensory inputs and intuitive system reactions when people use practices like mindfulness meditation or behavior monitoring technologies. In this model, it is important to note that neither system represents pure “reason” or “emotion” as in Kahneman’s (2011) two-system theory and that neither system always produces superior results. Instead, each system has different strengths and is prone to its own type of mistakes. Finally, please note that specific ovals and arrows—especially those in the lower half of the diagram—are for illustration and not intended to represent all possible pathways of influence.

Evidence From Contemporary Philosophy of Consciousness

Philosopher of consciousness Elisabeth Pacherie (2006) suggested that what is known about both neuroscience and phenomenology fits best with a view of two temporally distinct systems. People have experiences at one level of awareness but interpret those experiences at another. The intuitive system (in Pacherie’s terms the “present” or “P-level”) allows experiences to have an immediate and direct effect on behavior, whereas the more temporally distant narrative system (“future” or “F-level”) produces language-based abstractions about experiences. Linguistic constructs are always a step removed from the things that they represent. For instance, the first author lives with an animal who his narrative system identifies as “Midnight” (his name) but also as a “miniature poodle” and, more broadly, a “dog”—constructs that refer to no particular thing but rather to a general class of things. Midnight is a bit high in the shoulders compared with expected norms for the breed yet not so tall that anyone would call him a “standard poodle” instead. Children learn this type of abstraction during normal development; it is a key function of language that lets people move from the particular to the general and back again. Abstraction also provides practical benefits: If I know only about Midnight, I have a limited frame of reference. However, if I know something general about poodles, dogs, or mammals, I can respond effectively to many scenarios that may never have arisen with this specific dog. Similarly, statements of intention exist in language and are temporally distant from the behaviors they reference, allowing for predictions but also for errors in judgment. For instance, I may have absolute confidence that I will be able to resist the temptation of cheesecake after dinner, but when the dessert appears, I may have an experience that I describe as a “failure of willpower.” Later, I may notice what happened and try to reconstruct some reasons for my bad behavior, a process that again occurs in language. In this example, I might not even have been strongly aware of my cheesecake-eating behavior when it occurred, a phenomenon in line with evidence that human consciousness is “sparse.” Attention is involved in decision-making only when people deliberately focus on goals, and it is a scarce resource easily drained by the competing demands of life (Schwitzgebel, 2011). Some studies even suggest that attention depends on plasma blood glucose, which becomes depleted with effort resulting in worse decisions (Gailliot et al., 2007). Philosophers of consciousness suggest that the intention–behavior gap occurs simply because all narrative models of experiences and behaviors are a step away from the actual production of behavior in context.

Evidence From Cognitive Neuroscience

Two-systemmodels are well established in cognitive neuroscience (Evans, 2008). Intuitive processing is linked to activity in the mesolimbic dopamine system including the nucleus accumbens and striatum, whereas narrative operations are linked to activity in the lateral PFC (McClure & Bickel, 2014). This distinction fits with Plato’s original hypothesis that the intuitive system is more “emotional,” although the limbic system corresponds more exactly to the concepts of arousal or stress, with emotional labels attached by the narrative system after the fact (Sanfey & Chang, 2008). Consistent with the intuitive system strengths described above, subcortical areas have faster response times (Evans & Stanovich, 2013) and greater activation during intuitive states like meditation or jazz improvisation (Paulson, 2013) and produce better results than the PFC during highly practiced behaviors like a golf swing (Lisman & Sternberg, 2013). For people’s other mode of thinking, Evans and Stanovich (2013) suggested that the dorsolateral PFC is both strong in abstract thinking and involved in inhibiting emotional responses, two characteristics attributed to the narrative system. Multiple reviews suggest that the PFC performs symbolic representations to generate controlled actions, while intuitive system decisions are made using heuristic rules: Sanfey and Chang (2008); McClure, Laibson, Loewenstein, and Cohen (2004); and Koenigs and Tranel (2008) each found that logic-based responses were linked to cortical activation, whereas faster heuristic-based ones originated in the limbic system. In the study by Koenigs and Tranel, these decision-making differences disappeared among participants with lesions in the ventromedial PFC. Sanfey and Chang found that whether a logical but unfair offer was subsequently accepted or rejected could be predicted by activation differences between the dorsolateral PFC (associated with selecting a logical but unfair outcome) and the anterior insula (associated with reacting based on fairness). Not only are intuitive system decisions independent of conscious awareness, they are frequently not even stored in memory, as in many people’s experience of not remembering their drive to work (Barrett, Tugade, & Engle, 2004; Lisman & Sternberg, 2013; Squire & Dede, 2015). Blalock and Reyna (2016) summarized evidence on health decision-making to show that people often make one type of decision based on the gist of health information presented (narrative system) but another based on specific details (intuitive system). Although the evidence does not completely rule out a role for the narrative system in producing behavior, the data as a whole support a model in which the midbrain intuitive system processes stimuli and generates behavior, whereas the cortical narrative system comments on behaviors and their consequences after the fact. Although the PFC is often identified as the seat of reason or executive control, several reviews suggested that its role may be more connected to imagination—the ability to visualize oneself in the future (Bickel, Moody, Quisenberry, Ramey, & Sheffer, 2014; Stanovich, 2009). Hancock (2015) argued that the PFC is primarily a “possible world generator” as shown by its strongest levels of activation when people are dreaming. In line with Stanovich’s (2009) ideas, we suggest that the narrative system uses language, imagery, and social imagination for “safe experimentation” with possible courses of action similar to methods used to shape behavior in artificial intelligence studies (Amodei et al., 2016).

Behavioral Economic Findings

Kahneman’s groundbreaking studies of bounded rationality are remarkable precisely because they found that people’s in-the-moment decisions are quite different from their stated preferences at a more abstract level of thought (all examples from Kahneman & Tversky, 2000). For example, in well-known “endowment effect” studies, people had no initial preference between a $5 mug and a $5 bill, but after receiving one of the items, they were unlikely to trade it for the other. In our view, these findings are simply evidence of the differences between intuitive and narrative thinking. Language-based reflection is engaged by the hypothetical choice between a mug or currency and considers them of equal value in the future, but with one of the items in hand, the intuitive system says people should keep what is theirs in the present. Similarly, people often select lower-risk medical treatments, choosing drugs with fewer side effects but less efficacy (intuitive goals in the present) over those with greater benefits but a higher risk (narrative considerations among people who are not currently ill). Likewise, there are major differences between people’s willingness to accept environmental health threats (e.g., you have been exposed to a toxin; what should the company responsible offer in damages?) and what they would pay to avoid the same situation (e.g., what is it worth to keep Company X from dumping toxins in the park near your house?). In this scenario, people thinking intuitively about present exposure will sue for more money than they are willing to pay, on reflection, to avoid future exposure. Similarly, a recent nursing study found that older adults think home-based health sensors are a great idea for other people or in the future (narrative thinking) but too inconvenient for their own current use (intuitive thinking; Reeder et al., 2013). Behavioral economists have proposed long lists of cognitive biases to account for such findings, but we argue that each can be parsimoniously explained by temporal immediacy: the framing of a question as present versus future focused.

Evidence From Nursing Symptom Science

The principle of temporal immediacy can also elegantly account for experiencing/remembering discrepancies in people’s ability to cope with chronic diseases (Riis et al., 2005). Although most people predict that negative health events will strongly diminish their quality of life, they tend to be more resilient than expected when such events actually occur (Ubel, Loewenstein, Schwarz, & Smith, 2005). Similar issues plague end-of-life care, where people’s expressed preferences in the moment (intuitive) may be different from what they once expected themselves to prefer (narrative; Ditto & Hawkins, 2005). Symptom reports are also sensitive to item wording that elicits an intuitive or narrative response. For instance, when people with chronic obstructive pulmonary disease received instructions that prompted intuitive responses via concrete imagery (e.g., “imagine blowing into a tube with a ball”), they rated the same level of air resistance as less distressing than when asked in a way that triggered narratives (e.g., “imagine you are breathing as you did during your clearest memory of breathlessness”; Meek, 2000). Similarly, pregnant women rated their pain much higher in the immediacy of early labor (intuitive system) than when asked to reflect on their experience of early labor pain during the postpartum period as part of a narrative about their entire labor and birth experiences (Lowe & Roberts, 1988). These are endowment and anchoring effects in behavioral economics terminology, but we again suggest that temporal immediacy is sufficient to explain them.

Ecological Momentary Assessment

Finally, the principle of temporal immediacy explains findings from ecological momentary assessment studies that show substantial differences between thoughts, feelings, and behaviors measured in the moment using technology and the same constructs measured in aggregate after the fact. Such differences have been found for people’s reports of coping behavior (Ptacek, Smith, Espe, & Rafffety, 1994), mood (Jahng, Wood, & Trull, 2008), pain (Stone, Broderick, Shiffman, & Schwartz, 2004), and motivation to take medication (Cook, Schmiege, Starr, Carrington, & Bradley-Springer, 2017). Schwartz, Neale, Marco, Shiffman, and Stone (1999) argued that there is no such entity as “trait coping” because measures of coping in the moment are more variable within persons than between persons, and a person’s self-reported “usual” coping accounts for less than 10% of the variance in his or her daily coping scores. Yet, researchers and laypeople do talk about coping as though it were a stable disposition—a trait rather than a state. This again is explained by the difference between people’s narratives about coping strategies and the intuitive processes that produce actual coping behaviors.

Synthesis

We have presented historical and contemporary evidence that the greatest challenge in health behavior change is the intention–behavior gap. We then presented Kahneman’s theory of mind as a potential advance over prior models, but one with important limitations: It prioritizes narrative over intuitive thought, and it does not have a clear neuropsychological mechanism for predicting whether one system or the other will be activated. We then proposed an organizing principle—temporal immediacy—to explain whether the intuitive system or the narrative system is activated in any given research context. The idea of temporal immediacy is based on our argument that the intuitive system governs all behaviors in the present whereas the narrative system is only useful in modeling the past or the future. Finally, we reviewed five different knowledge domains and presented evidence from each that is consistent with our elaboration of Kahneman’s model. In the final sections below, we explore research and clinical implications of our views.

RECOMMENDATIONS FOR RESEARCH

Theory

First, the time is right to reevaluate traditional theories of health behavior. This is not only because of recent developments in neuroscience and behavioral economics reviewed above but also because new technologies have dramatically increased our ability to study everyday experiences. This window into intuitive processes has not existed previously. Daily experience surveys are equally effective whether administered on paper or electronically (Tennen, Affleck, Coyne, Larsen, & DeLongis, 2006), but technology facilitates methodological features like random cues and collection of collateral environmental data. Such data are essential for improving our understanding of intuitive mental processes, and the findings of these studies are likely to show different relationships between variables than those seen in well-established models based on narrative level data (Riley et al., 2011). Because technologies that capture people’s experiences as they occur are relatively recent, the theory basis of studies using these technologies is still evolving, with weak or absent theoretical frameworks and analyses driven mainly by the data available. Better models of intuitive thinking are urgently needed to flesh out the details in Figure 1: What cognitive processes operate at the intuitive level, at what points does the intuitive system consider narrative input, and how does the intuitive system select specific behaviors? Future models should consider neurocognitive processes as well as behaviors based on longitudinal observation and the specific contexts in which health behaviors occur.

Design

Second, a recent panel of eminent nurse researchers (Corwin et al., 2014) recommended strategies for symptom science that are consistent with our theory, including the suggestion that symptoms should be studied and managed in context—as intuitive processes in the circumstances where the symptom actually occurs. Important contextual factors include stressors, interpersonal support, work effort, leisure time, mood, anxiety, headache, and sleep. All of these variables can be studied in context using ecological momentary assessment. New technologies are available to capture self-report and biological data together, like a rating scale and heart rate variability as simultaneous measures of stress or survey-based and actigraphy data as simultaneous measures of sleep quality (Corwin et al., 2012). Longitudinal designs using intensive within-person data collection are essential, and using multiple measures in context will help researchers better understand the intuitive system, including links between physiology, behavior, and symptom self-reports.

Measures

Third, we recommend novel measurement strategies to study intuitive thinking. It seems likely that narrative processes can gain access to at least one intuitive-level experience at a time through focused attention. Some meditation practices are designed to do exactly this: help people attend to experience while avoiding language-based judgments about it (Paulson et al., 2013). New technologies make it possible to gather ongoing ambulatory data on parameters like heart rate variability as an indicator of stress, skin conductance as a measure of mood, geolocation data on environmental risks, activity or inactivity, sleep quality, ambient light or sound, or even brain activity as an indicator of attention (Reeder, Cook, Meek, & Ozkaynak, 2017). Furthermore, these devices may support the same goals as mindfulness meditation if they help people to build more accurate narratives about their intuitive system experiences. Philosopher of consciousness David Chalmers (1997) suggested a difference between registrations, which are the perceptual perceptual inputs available to the intuitive system, and judgments (the subset of registrations that people notice). Through focused attention, one can become consciously aware of many registrations, such as thoughts, feelings, sensations, symptoms, and even autonomic processes like heartbeat, although it is likely that some registrations are permanently outside awareness and nevertheless influence behavior (Schwitzgebel, 2011). For registrations that enter awareness, methodological features like random cues and concrete questions help to avoid accidental contamination by narrative level beliefs (Bearinger et al., 2011). Including variables like mindfulness in studies using narrative methods could also help to improve the accuracy of behavioral predictions (Chatzisarantis & Hagger, 2007). Technology also allows unobtrusive monitoring of behaviors or physiological states outside language. Our view of symptoms as inherently biobehavioral (Corwin et al., 2012) means that they should be measured through physiological indicators as well as via self-report. Data sets with both physiological and psychosocial measures are likely to give the best picture of patients’ immediate experiences or behaviors. A recent example of research using intuitive system data is HeartMath, an approach to moving the mind from chaos to coherence that measures heart rate, breathing, and blood pressure (Edwards, 2015).

Analysis

Fourth, advanced modeling strategies are needed to analyze the intensive within-person, longitudinal data collected through measures of daily experience (Bolger & Laurenceau, 2013; Schwartz & Stone, 2007). Multilevel modeling reflects a broad class of analytic techniques, including longitudinal growth curve modeling to address multiple time points nested within individuals and dyadic analysis to address multiple indicators within individuals (e.g., sleep diary and actigraphy). Multilevel models can be used to quantify within- versus between-person variability in daily experience studies and to characterize changes in intuitive processes over time (Walls, Jung, & Schwartz, 2006). Multilevel modeling also has flexibility to address various aspects of study design, such as multiple time points (including differences in available time points across participants), different levels of assessment (e.g., between and within person), and various roles assigned to study measures (e.g., mediators, outcomes). Because trajectories may not be linear or smooth, it is important to consider nonlinear approaches that best capture the functional form of trajectory shapes over time, such as generalized estimating equations and nonlinear mixed models (Henly, Wyman, & Findorff, 2011). In addition, person-centered methodologies like growth mixture modeling may be used to determine whether there is population heterogeneity in the trajectory shape and to identify subpopulations that may differ from one another in terms of the direction or functional shape of their individual trajectories over time (Muthén & Muthén, 2000). Finally, dynamical systems analysis (Boker, 2015) is a novel method that uses differential equations to identify unobserved equilibria based on the study of variations over time; this method may be particularly useful in understanding the intuitive system’s decision-making heuristics and processes for selecting behaviors.

IMPLICATIONS FOR PRACTICE AND HEALTH BEHAVIOR INTERVENTIONS

Rejecting Platonic antagonism between the two human decision-making systems and a related preference for narrative thinking is difficult but allows for a view of the two systems as potential partners in producing desired results. Table 3 presents examples of how our theory might inform practice. Escaping from Plato’s good/bad, reason/emotion, and divine/base dichotomies is likely to take significant effort given their foundational status in Western thought. However, if the narrative system does not function “in the moment” and thus exercises no direct control over behavior, we suggest that efforts to use logic and reason as the primary means of health behavior change are doomed to fail. This is particularly true in contexts of information overload (Hwang & Lin, 1999), personal threat (Burke et al., 2010; Hennes et al., 2012), authoritarianism (Bocchiaro & Zimbardo, 2017), or competition (Burke et al., 2010; Puurtinen et al., 2015), where the narrative system is more likely to ignore observations and logical principles, in order to reduce fear or optimize personal benefits. In these contexts, intuitive reasoning can produce more beneficial results (Bocchiaro & Zimbardo, 2017). By recognizing that the conscious “I” is not a decision-making executive but a storyteller about past and future, clinicians also can better utilize the narrative system’s strengths by asking people to envision choices and their consequences.

TABLE 3.

Potential Interventions to Apply the Theory in Practice

Focus of intervention	Possible clinical strategies
Trick the intuitive system to select healthy options	Use behavioral-economic “nudge” strategies: for instance, make healthy choices more visually appealing or make the default choice the healthy choice and require a specific “opt-out” decision to make an unhealthy choice.a Associate with healthy peers, capitalizing on the intuitive system’s tendency to affiliate with others.b
Train the intuitive system to make better decisions	Create feedback to make the long-term consequences of behavior more immediate: e.g., “if you exercised every day like today, you would weigh X pounds in 1 month.” Use technology to prompt healthy behaviors—audio alarms, haptics, etc.c  Train people to attend to their present experiences, to increase awareness of intuitive processes and provide more opportunity for narrative reflection.d Practice healthy behaviors so that they become habitual. Create cues, associate them in memory with healthy choices, and then present these cues in the environment.  The intuitive system will be aware of them and modify behavior accordingly, even without conscious awareness of the cues.e Avoid creating cues that do the opposite: For example, a public service announcement that says “90% of college students drink too much alcohol” tells the intuitive system that “drinking too much is perfectly normals.”
Engage the intuitive system for unexpected solutions	Raise awareness of situations where the narrative system is likely to make leaps in logic or engage in biased reasoning, such as conditions of information overload,f in-group/out-group conflict,b authoritarianism,g or anxiety.h,i Restructure situational demands to avoid these problems when possible, e.g., reduce information overload by offering only a few salient points at one time. Teach people introspection or mindfulness strategies to increase their awareness and accuracy in reporting present experiencej or mindfulness skills to produce more accurate models of future behavior.k Elicit people’s positive intuitions about their health behaviors, which may serve as natural motivators for change. This requires recognition that, in cases of ambivalence, the intuitive system can simultaneously hold two opposing views.l Encourage “natural solutions” that build on people’s past experiences, current behaviors, and existing strengths.m
Engage the narrative system more effectively	Avoid relying primarily on facts and logic to convince people, based on evidence that this approach not only fails but also produces more resistance.l Encourage people to envision the social consequences of healthy behaviors or unhealthy ones, using the narrative system’s capacity for social imagination.n Encourage people to explore alternatives using vivid imagery, based on evidence that narrative thinking is especially connected to visual processing.o Encourage detailed elaboration and exploration of alternatives, e.g., by asking people for implementation intentions that specify when, where, and how they will undertake behaviors.p Provide tools, resources, or training to encourage creative and divergent thinking. Creative solutions most often emerge from a two-step process with generation of many different ideas (intuitive system) followed by an evaluation of appropriateness (narrative system), with synchronous activation of both brain areas.q

^aHalpern, Ubel, and Asch (2007).

^bPuurtinen, Heap, and Mappes (2015).

^cReeder et al. (2017).

^dPaulson et al. (2013).

^eCialdini et al. (2006).

^fHwang and Lin (1999).

^gBocchiaro and Zimbardo (2017).

^hBurke, Martens, and Faucher (2010).

ⁱHennes, Nam, Stern, and Jost (2012).

^jSchwitzgebel (2004).

^kChatzisarantis and Hagger (2007).

^lMiller and Rollnick (2002).

^mO’Hanlon and Beadle (1997).

ⁿBickel et al. (2014).

^oStanovich (2009).

^pGollwitzer (1999).

^qBeaty, Benedek, Kaufman, and Silvia (2015).

It may be possible to improve the intuitive system’s effectiveness through practice or using methods like mindfulness (Paulson et al., 2013) or behavior monitoring technologies (Reeder & David, 2016). Because the intuitive system learns from experience, researchers could design feedback interventions to develop unconscious expertise in health behavior domains. Furthermore, more accurate awareness of in-the-moment experiences might lead to better congruence between one’s intentions and one’s behaviors (Chatzisarantis & Hagger, 2007). Self-monitoring technologies may have a similar benefit if a greater awareness of what one eats, how much one exercises, or how one sleeps informs a new narrative to affect future behavior. The theory presented here might provide a stronger scientific foundation for mindfulness or other interventions focused on attention, brain activation, and states of consciousness.

In addition, intuitive thinking can sometimes identify solutions that the narrative system cannot put into words (Lisman & Sternberg, 2013). Zimmerman (2016) noted that most behavioral economic interventions do not use the intuitive system’s strengths but instead capitalize on blind spots to trick the intuitive system into better decisions. That may be why “nudge” interventions—subtle environmental cues like default options or social comparisons—can provoke visceral opposition from some people despite their demonstrated efficacy (Cialdini et al., 2006; Halpern et al., 2007). An alternate strategy would be to approach the intuitive system as a source of wisdom or potential solutions. For instance, Paulson et al. (2013) recommended meditation as a way to reduce distress by focusing on present experiences without adding narrative judgments. Alternately, researchers could teach people techniques for introspection (Schwitzgebel, 2011) or mindfulness (Paulson et al., 2013) to improve the accuracy and number of options that are available for intuitive decision-making.

Finally, involving both systems in decision-making may lead to better outcomes. Creativity is linked to synchronous activation of brain areas connected to both intuitive and narrative thought, with the data suggesting a two-stage creative process in which different brain responses are associated with the generation of novel ideas and subsequent judgments about the new ideas’ appropriateness (Beaty et al., 2015). Many meditation practices have an attentional component, suggesting narrative processes, yet their practitioners often describe them as ways to get beyond the “self” and in touch with a broader and more relational mind—an idea that seems more in line with intuitive thought. As Schön (1987) suggested, practice reflecting on one’s actions as they occur builds capacity for future reflection-in-action, leading to different behaviors by way of better narratives. Narrative and intuitive processes work together in this last category of change strategies, but it is important to keep in mind Stanovich’s (2009) suggestion that the narrative system is primarily a generator of ideas for the intuitive system to select and implement, rather than a final arbiter of decisions. Researchers should find new ways to encourage two-system solutions that involve “transformation” or a sudden switch to different results, rather than “change” based on rational argument and effort.

Conclusion

Implementing research and practice interventions that capitalize on the strengths of both systems will require continuous challenges to people’s Platonic assumption that the conscious mind makes decisions while intuitive processes are either mere noise or an actively oppositional force. Viewing behavior through the lens of temporal immediacy also might help people to assign less blame when failures of rationality or willpower occur. To make progress, people need strategies to enroll both systems in support of their goals. By keeping in mind the idea that the narrative system has no direct causal role in producing behavior, people may be less surprised at how challenging it is to change health behaviors and could become more successful in addressing this longstanding problem.