Damian Sendler: The following is an outline of the current article. Three prominent experimental paradigms for introducing conflict will be described, beginning with a brief introduction on how the concept of “conflict” has been considered in the history of psychology. To create conflict in these paradigms, stimuli are presented that, in addition to being processed in accordance with instructions, elicit additional cognitive processes that interfere with the instruction-based processing of data. Contrary to expectations, these conflict-inducing tasks have been found to be capable of inducing conflict at multiple levels of information processing at the same time. This raises a number of interpretive issues. After that, we’ll go over some key theoretical ideas. After that, we’ll take a look at how these tasks have been used in clinical settings. Pathological changes in information processing can be detected through comparisons between clinical and control groups, or by tracking the effects of therapy on an improved task performance that is more similar to that of non-pathological participants. These pathological changes can be detected through comparisons between clinical and control groups. These conflict tasks, however, often undergo structural changes that alter their underlying processing requirements, sometimes making them completely different tasks despite their superficial similarity to their original prototypes, when they are transferred to a clinical context. The article concludes with some thoughts on the mechanisms that may underlie certain patterns of performance of these tasks, as demonstrated by clinical populations.
Damian Jacob Sendler: Since the beginning of the twentieth century, the concept of conflict has permeated nearly every branch of modern psychology. Some argue that resolving conflict serves as an important conceptual engine for the transformation of intangible (and thus unobservable) entities into observable behavior. If, for example, the Ego brings psychological forces into contact with reality, it is thought to arbitrate conflicts between biological drives (the Id) and cultural norms (the ego) (the Super-Ego). By replacing earlier responses with new ones, Ach [1] created the first conflict-inducing experimentation paradigm, which was pioneered in the study of human conflict resolution by Ach [1]. The idea of conflict between multiple conditioned or unconditioned responses has been used by behavioral psychologists to explain complex overt behavior (e.g., ‘conflict theories’ of ‘experimental neurosis’).
Dr. Sendler: According to Lewin’s typology [3,] motivational conflicts have been a major influence since the early 20th century. It was Lewin who made a distinction between three basic types of conflict: the conflict between two approaches (the conflict between two approaches being avoided, the conflict between two approaches being avoided) and the conflict between two alternatives that results in the conflict of both approaches (the conflict of two approaches being avoided). Many clinical problems, such as anxiety, have been influenced by Lewin’s work and subsequent work by Miller [4] on the steepness of avoidance- and approach-gradients. Gray’s Behavioral Inhibition System theory has also been applied (e.g., [6]).
A more mechanistic flavor to conflict was introduced by cognitive psychology, which drew on metaphors from information technology (e.g., interference, cross-talk). More experimental paradigms were developed that used conflict in a more controlled way. According to early selection accounts of attention, the first experimental paradigms viewed conflict primarily as a conflict between various sources of information (e.g., dichotic listening). With the rise of late-selection accounts, there was an increase in conflict on the response side. Even seemingly simple conflict tasks (which were originally thought to capture conflict on either the stimulus or response side) can now induce conflicts on multiple levels at once, as has been demonstrated in recent years. Experimental conflict tasks have produced a wealth of new information about human information processing over the last few decades, but they have revealed a number of issues and dangers that should not be taken at face value. Also established is that tasks are represented on multiple levels that are not adequately characterized by the traditional distinction between perceptual, central and motor levels. There is a corresponding increase in the number of potential conflicts that can arise.
Even the simplest of tasks can spawn a slew of conflicting processes, as will be demonstrated in greater detail below. There is no guarantee that even further simplifying tasks will make the conflicts they cause more visible. If you restrict the number of stimuli presented, for example, you may open the door to new conflict types that would otherwise go unnoticed with a larger stimulus set, for example. To illustrate this, consider the Eriksen flanker task, in which participants must respond to a central target while ignoring distractors that flank the target. If the current trial is preceded by an incongruent trial, as opposed to a preceding congruent trial, the current trial will be more efficient at a target with incongruent distractors. Using a smaller versus a larger stimulus set, Bugg [7] demonstrated that the conflict adaptation effect in the flanker task is masked by additional and counteracting negative priming (an inhibition of a previously ignored stimulus attribute), implying that a more complex task allows for a more pure measure of conflict adaptation than a less complex task.
A stimulus-response conflict task has a relevant stimulus-response that maps the implementation and can be interfered with by a nominally irrelevant stimulus or stimulus attribute. These three types of conflict tasks are introduced in the following. A competing response is elicited when an irrelevant stimulus attribute shares a conceptual dimension with the relevant stimulus attribute, as in the ‘classical’ versions of these tasks (e.g., the meaning of a color word whose color is to be categorized, or stimuli that surround a target of the same type). Ensembles of type 2 or type 4a or type 4b according to the taxonomy of Kornblum, Hasbroucq, and Osman [7]. They may change their taxonomic status after discussing clinical versions of these tasks, in which the nominally irrelevant stimulus attributes draw their potential for interfering with the relevant stimulus-reaction mapping from other sources (as will be discussed below). The Implicit Association Test [9], affective priming tasks, or approach-avoidance tasks are explicitly excluded from this section because they primarily generate conflict on the basis of evaluative responses (cf. [10]). In order to avoid adding unnecessary steps, the evaluations incorporated into the design of these tasks must be restricted (which is also the case with affective variants of the Simon task, cf. [11]). The implicitness of measurement procedures and measurement outcomes are also addressed by these paradigms [12]. This article has no business delving into the depths of these theoretical issues.
The Stroop task is one of the more traditional conflict resolution methods in psychology research, despite its age. Four conditions were presented via stimulus cards in the original version of this task [13]. Two different conditions were used to test participants’ reading abilities: one required them to read words printed in colors that did not match the words they were asked to read (e.g., the word red was printed in green), while the other required them to read the same words printed in black. However, in a different experiment, participants were asked to identify the color of words that denoted incongruent colors (basically identical stimuli as those used earlier), while those taking part were asked to identify the color of patches instead. Color words and non-word stimuli have been compared before, but Stroop’s ingenious move was to combine words and colors into two-dimensional stimuli that could be responded to according to different instructions for word-reading and color-naming. While reading color words printed in black wasn’t affected by the incongruity of the ink colors, color naming was severely hindered by incongruity of the ink colors (as opposed to reading color words printed in white) (as compared to naming the ink color of patches).
The term ‘Stroop-like interference’ refers to a variety of two-dimensional (and possibly conflicting) stimuli that have been shown to elicit (at least superficially) similar interference. Original Stroop-interference has long been attributed to a difference in practice that made word reading ‘automatic’, while color naming was viewed as “voluntary” [14]. It has been widely debated since the 1960s and onward, but the automatic-voluntary distinction remains perhaps the most commonly used shorthand account of Stroop-like interference phenomena. These facts hold even though it is widely acknowledged that in recent years, the automatic-voluntary distinction in cognitive psychology has lost almost all of its credibility [17]. (at least when considered in terms of a dichotomy).
Stroop-like interference can’t be reduced to two conflicting processes (automatic and voluntary), but at least three types of conflict: semantic conflict, task conflict and response conflict. In addition, these conflicts are subject to a variety of cognitive control mechanisms (e.g., proactive and reactive, cf. [19]). A thorough task analysis is required to determine which specific conflict-control configuration is activated by a given experimental protocol in the universe of Stroop-like tasks because there is such a wide variety of conflicts and control mechanisms.
For the purpose of studying attentional processing, Eriksen and Eriksen [20] developed the flanker task. Three identical noise letters on either side of the central letter were used to elicit a response in the game’s original form. Pressing a left or right lever assigned two letters to each response was the only method of completing the task (H and K vs. F and C). There were three main experimental conditions: the noise letters (flankers) could either be identical to the central target letter, differ from the target letter but belong to the same response set (compatible condition), or belong to the other response set (incompatible condition). Major findings included a marked slowing of responses in the third condition compared to the first two, a phenomenon dubbed “flanker interference” later. Although processes related to the spatial distribution of attention play a role, there is consensus that flanker interference is primarily due to response competition, as it has been shown that the incompatible flanker condition results in a (mostly transient) activation of the incorrect alternative response [22].
Variations of the original flanker paradigm exist in terms of the number and types of stimuli, spatial layouts of targets and distractors, as well as the (a)synchronicity with which targets and flankers are presented. Examples of these variations include: In contrast to Stroop-like tasks, most variants involve processing flankers and targets in the same task set. Targets and flankers in the flanker task are of the same type and undergo the same encoding operations as targets and flankers in Stroop-like tasks (e.g., color categorization vs. word reading, causing conflict on the task-set level). Therefore, the flanker paradigm does not have a lot of room for task conflict. If the flanker stimuli used are only semantically related to the targets, this feature is lost [23], which is a common problem with this type of flanker task.
For example, a go signal and a stop signal can be put against each other in a methodologically elegant way that has yielded important conceptual insight into the nature of ‘automaticity,’ which is discussed further down. As a result, it has been included.
In the 1940s, Vince [24] developed a task that required participants to track the displacement of a horizontal line with a pointer, which was the first use of a stop-signal task. Inter-stimulus intervals (ISIs) ranged from 50 to 1600 ms on some trials, with shorter intervals meaning that participants had to stop a movement in the direction of the first displacement and perform an opposite movement on the following trial. According to Vince, responses to the second stimulus were delayed more than expected when ISIs of 500 milliseconds or less were used in his experiment. It was concluded that after the first stimulus was presented, “the sensori-motor system is refractory to similar stimuli,” resulting in an additional delay of 500 ms or less for ISI measurements. The concept of inhibition and conflict are not mentioned in this account.
Lappin and Eriksen [25] published an experiment about 20 years later that included most of the elements of contemporary stop-signal tasks, including the concept of’response inhibition.'” Participants were instructed to react only if a second light with an ISI of 0, 12, 33, or 63 ms was not shown after the first. It was considered “competitive” to process the first light as a “go signal” and the second light as a “stop signal.” Individual mean RT to go trials and the duration of the ISI separating the onset of the go and stop signals in stop trials were found to be jointly determinative of the probability of successfully stopping the response to the go signal in the case of the presentation of a stop signal, according to the authors.
This paradigm, which was developed more than two decades later by Logan and Cowan [17], serves as a model for the vast majority of modern stop-signal tasks. All participants had to choose one of four letters, each of which was associated with a 2:1 response mapping. Go signals were followed by auditory stop signals ranging from 50 to 500 ms in about 25 percent of the trials Because there is no obvious response to a stop signal, it is impossible to observe the latency of the stop-signal reaction time (SSRT), which Logan and Cowan calculated using the assumption of a race between a go and a stop process that was considered stochastically independent.
It is important to note that the stop-signal task relies on the fulfillment of the race model’s basic assumptions to estimate SSRT, which makes it vulnerable to any violations of these assumptions. Short stop signal delays are required when using a very easy go task with fast response times to allow for participant stop signals. It’s possible that this will cause perceptual interference between the processing of go and stop signals, which would violate the assumption that the go and stop processes are separate [26]. There is a similarity between the effects of variable foreperiods and the effects of stop signals on participants’ expectations and strategies, which can have a significant impact on results [27].
The stop-signal task, in conjunction with its formal underpinnings, enables an empirical determination of degrees of automaticity (as opposed to control) that are otherwise only assumed in many cases. For instance, many people believe that ‘automatic’ processes are inherently dangerous (that is, unstoppable). It is possible to determine the existence of a ballistic processing component empirically through the estimation of SSRT, because there should be a flat inhibition function for the duration of the ballistic component.
Damian Sendler
Researchers have found that having a subject press the right key in response to left-pointing signals while pressing the left key in response to right-pointing signals activates response tendencies that lead to a temporary activation of the incorrect but spatially compatible response. To describe this as an automatic action is to say that it goes against the user’s intent to press the instructed key and requires the user to exert cognitive control to avoid errors. To be clear, it is obvious that the same participant will not hit any key on their right side when they see an arrow pointing in the right direction while they are outside of the lab. An examination of this brief sketch should cast doubt on the assumption that stimuli are endowed with power to elicit a response independent of intention or attention—or that there is completely stimulus-driven automatic behavior.
On the other hand, the concept of automaticity rather than control is used in nearly all accounts of conflict-induced interference (and facilitation). Rather than seeing automaticity primarily as fast processing due to extensive practice, later theories viewed it as a special case of information processing (and behavior) that was not only characterized by its speed but also by its independence from intention and awareness, its independence from limited attentional resources, and its ballistic nature. (e.g., [28,29]). Monolithic automaticity dominated cognitive psychology in the 1970s and 1980s, but was replaced by various variants of ‘conditional’ or ‘prepared’ automaticity that, in addition to dissociating various features of monolithic automaticity, argued that there is almost no kind of information processing and behavior completely independent of intention, proceeds outside of awareness, and unstoppable. [30]. (e.g., [17,31]). Even more significant than simply denying that all behavior is either controlled or uncontrolled, abandoning monolithic automaticity also ruled out any presuppositions that one feature of automaticity necessitates the other features (e.g., that ballistic behavior is unintentional and runs off outside of awareness).
Due to the lack of observable behaviors that are ‘process-pure,’ or that reflect only one underlying process regardless of whether it meets any of the criteria for ‘automaticity,’ the concept of automaticity can be problematic. Overt behavior (or its absence due to successful stopping) may be based on a number of competing or synergistic processes, some of which can only be inferred from overt behavior. As one of the most popular solutions, two distinct paths of information processing are proposed that eventually lead to the same outcome.
As far as cognitive psychology goes, the Dimensional Overlap (DO) model [8] is the most widely known. Modeling a wide range of stimulus-response compatibility effects is the primary goal of this model. Stimulus-response compatibility is typically based on dimensional overlap between stimuli and responses, according to the DO model. In tasks of the Simon type [32], participants are asked to respond to a non-spatial attribute of a stimulus that is presented in different spatial locations with a spatial response that varies on the same spatial dimension as the stimulus locations, and dimensional overlap is most obvious when stimuli and responses vary on the same dimension (e.g., left and right stimuli are responded to by left and right key presses). There are two methods of information processing in the DO model: activation and confirmation. No matter if instructions state otherwise, stimuli and responses are assumed to activate the same response regardless of whether this is the correct response. This activated response is confirmed and executed immediately in the case of a compatible mapping of stimuli to responses, whereas in the case of an incompatible mapping, this activated response must be transformed or replaced by the appropriate response, resulting in a slower response. In two ways, the DO model assumes that activation is automatic in the DO model: First, it is assumed that recoding and transformation occur “without any interference or intervention by monitoring and controlling processes” during processing stages (e.g. stimulus identification) ([8], p. 262). To begin with, it is presumed that the output of one stage, such as stimulus identification, is transmitted directly and “without interference or intervention” to subsequent stages (such as response selection) (ibd.). [8] on the other hand, consider the concept of automaticity to be lacking “in adequate detail and with sufficient confidence about the automatic processes in different tasks to provide much insight at that level” (ibd.). As a result, this dual-route structure has been used in other conflict tasks, such as the Stroop and Eriksen types of conflict tasks.
Damian Jacob Markiewicz Sendler: Until the 1990s, response-time research in cognitive psychology (including the conflict-related research discussed here) treated single experimental trials as quasi-natural units of analysis representative of behavior in general, apart from a few notable exceptions (e.g., [33,34,35,36]. There has been a general practice, as stated by Broadbent ([37], p. 876], to deliver a stimulus, catch a response, and ignore what occurred before and after). Because of this newfound interest in sequential effects, researchers have been investigating how one trial’s results are influenced by its predecessors. There has always been a role for sequential effects in conflict tasks; they had previously been overlooked, but now they are the focus of an enormous amount of research. Research on task switching and conflict adaptation are two of the most common approaches.
As previously discussed, research on task switching (cf. [38] for a review) introduces yet another type of conflict. To put it simply, the conflict is between continuing the work that has already been done and starting a new project altogether. An example of cognitive or behavioral perseverance can be found here Neuropsychologists have long studied perseveration using the Wisconsin Card Sorting Test [39], but recent methodological advances in task switching research have allowed for much finer-grained analyses of this behavior.
Damian Jacob Sendler
The sequential congruency effect is probably the most important sequential effect when it comes to conflict task performance. The Eriksen flanker task [40] was the first to show that interference effects are reduced after a second incongruent trial, as compared to a congruent or neutral predecessor trial. It has since been observed in all of the conflict tasks discussed in this article (cf. [41] for review). High trait anxiety has been shown to impair conflict adaptation [42], while evidence for other disorders, such as schizophrenia [43] and depression [44], is conflicting at best. The fact that experimental studies into the conflict adaptation effect have only recently become a focus of attention may have something to do with the fact that the underlying mechanisms of the effect are still hotly debated. Some argue that this effect is actually a result of cognitive control, while others argue that it’s simply the result of basic memory processes [45].
The conflict-monitoring theory [46] is the most widely accepted explanation for the sequential congruency effect. Thus, researchers have been forced to consider not only the characteristics of the most recent trial (trial n), but also those of the previous trial (trial n 1). Thus, clinically relevant observations in conflict tasks can now distinguish between changes in conflict monitoring (mainly as effects of trial n characteristics) and adaptations of cognitive control to the occurrence of conflict (mainly as effects of trial n 1 characteristics), significantly expanding the range of possible clinically relevant observations in conflict tasks. It is also necessary to control the relative frequency of relevant trial n features, as well as the transition frequencies from trial n 1 to trial n, as one further practical implication of this conclusion. Otherwise, one might interpret sequential effects as a result of trial n’s features [47], which could show that the majority of the ’emotional Stroop effect’ is a’slow’ one, affecting the trial following the presentation of a conflicting emotional stimulus.
Conflict tasks are frequently employed as diagnostic tools in clinical psychology [48]. On the one hand, it is possible to compare two conflict tasks: an affectively neutral task (like the Stroop color-naming task) and a ‘clinical’ version, in which conflicting information is linked to the clinical group’s pathology (e.g., the Stroop color-naming task) (e.g., an emotional Stroop task, in which distractor words are associated with threat in the case of anxiety as the relevant pathology). The clinical version of the conflict is not caused by an incongruent stimulus activating an incorrect response (saying “blue” instead of “green” in response to the presentation of “blue” in green color), but by the distracting stimulus feature, which compromises the processing of the response-relevant stimulus attribute (e.g., a slowdown of the verbal response “green” to the word “green” written in green). For example, it has been shown that women with anorexia and other eating disorders have more Stroop interference with food and body-related terms than with neutral terms (cf. [49] for review; cf. also [50]), that depressed participants have larger interference with emotional (positive but more-so negative) words than with neutral ones (cf. [51] for meta-analysis; cf. also [52]), or that the attentional control in a classical Stroop task suffers from the concurrency effect. It has been reported that a group with pathological worry proneness made less omission errors and more commission errors when the go signal was a worry-related word, as compared to the reverse condition, an asymmetry that was not shown in the control group [54]. It has been shown by [55] that when the target is a negative word, [56] there is greater facilitation by negative flankers and less interference by positive flankers when it comes to the flanker task. For example, high trait anxiety has been shown to compromise conflict adaptation [42], while evidence for other mental illnesses, such as schizophrenia [43] and depression [44], is more contradictory. In addition to being used for diagnostic purposes, clinical conflict tasks can be used to test hypotheses about the functional location of these effects, which could lead to a better understanding of the underlying pathology.
The Stroop task is the most commonly used conflict task in clinical research, but many of the same principles apply to other conflict tasks as well.
Interfering stimulus attributes have taken on a new significance with the introduction of “emotional” Stroop task variants and their subsequent use in clinical settings, as has already been stated. In the original Stroop task, the color of distractor words can be translated into a similar response, but employing emotional words associated with patients’ personal concerns alters the basic structure of this task.” No plausible candidate for causing conventional Stroop interference (semantic conflict, task conflict, response conflict; cf. [18]) can be identified when the distractor words (or associated key presses) do not activate color words (or associated responses). This suggests that there are likely to be other processes that play an important role in this.
A slow sequential process after the presentation of an emotional stimulus affects the trial in highly anxious participants, showing increased emotional Stroop interference, making it unlikely that attentional capture is the driving factor (cf. [57] for meta-analysis). As a result, attentional disengagement appears to be the problem. Evidence for pathology-related compromised attentional disengagement but unaffected engagement should be noted, however, because most studies using tasks requiring shifts in spatial attention may have been affected by a process of behavioral freezing, interfering with the engagement and subsequent disengagement of spatial attention, which may have been mistakenly attributed to delayed disengagement [58].] To put it another way, if you’re experiencing delayed attentional disengagement, it’s likely because you’re spending less time engaged in the task at hand.
Damien Sendler: The use of conflict tasks in the clinical setting is particularly appealing because they can accurately reflect the progress made in therapy [59]. Clinical studies looking at how therapy affects conflict processing promise to reveal mechanisms underlying pathology-induced shifts in clinical populations’ performance in conflict tasks, as well as potential targets for therapeutic interventions.
A dual-process theory contrasting “automatic” response tendencies with “goal-directed” controlled processing is commonly used to explain the decline in performance of conflict tasks in clinical populations when conflicting stimulus attributes are related to the patients’ pathology. Either the conflicting stimulus attribute captures the attention of patients, which is considered to be mostly automatic, or patients have difficulties in shielding and/or disengaging themselves from this information, with the relative importance of these factors possibly differing between different pathologies. Many studies have supported these claims, but it is clear that such dichotomies are an oversimplification of the situation. But in clinical practice, shorthand accounts may suffice if their limitations are taken into consideration.
However, it appears that additional, perhaps more significant, considerations have been largely overlooked. Examples include stimuli associated with patients’ personal concerns having a higher subjective frequency, which should make it easier for patients to retrieve the information associated with those concerns [60]. The assumption that differences in subjective frequency alone are the primary driving factor is challenged by the fact that psychotherapy usually includes activities related to the specific pathology, which should increase rather than decrease the frequency of exposure to pathology-related information [59]. When comparing two groups, the stimulus material used should be carefully chosen so that the subjective frequency of stimuli in each group is balanced [61]. A similar consideration should be made when determining whether or not stimuli’s valence (such as their aversiveness) varies between groups in order to avoid confusing stimulus valence with its relevance to individual concerns [61].
With regard to aversion, this may be especially relevant. There is a tendency to avoid aversive stimuli as the default response when confronted [62]. Implicit response conflicts may occur to different degrees when stimuli differ in subjective aversiveness between an individual in the clinical setting and an individual in the control group. Research has shown that conflict generates an aversive emotional response (cf. [63] for a review). Cognitive control processes are thought to be activated as a result of this aversive signal. The resource conservation principle [64] predicts, however, that people’s willingness to invest effort will drop if they consider the task to be exceedingly difficult because engaging in cognitive control is perceived as effortful and expenditure of effort is also perceived as aversive. In the implicit-affect model [65], cues associated with failure should make a task more difficult to perform. A strong association between pathology-related stimuli and failure can be expected in patients who suffer from specific pathologies; otherwise, they wouldn’t have the pathology anymore. That patients are less motivated to work on a task requiring cognitive control (e.g., inhibition) of conflicting information due to their pathology is therefore entirely conceivable.
A view like this may also be able to explain an observation made by [67]. Participants who were scared of snakes were given a Stroop task to complete, in which they used snake-related and neutral words. Participants had the option of completing this task with or without the assistance of a live boa constrictor. As a result of the snake’s presence, the phobics may have been more susceptible to snake-related words interfering with their thoughts. However, it was only in the absence of the snake that participants exhibited interference by snake-related words. They also reacted more quickly when the snake was nearby. Explain this by assuming that participants increased their effort under real threat, which also shifted subjects’ priorities away from the relatively mildly aversive snake-related words. Taking a closer look at the procedure ([67], Exp. 1) used in their snake-present condition leads to a slightly different interpretation: participants were exposed to a glass tank containing the snake first and then told that after performing the Stroop task, “they would be asked to do a behavioral test to determine how close they could come to touching the snake.
If they wanted to stop, “subjects were assured they could do so at any time” ([67] p. 521). As a result of this procedure, the participant’s autonomy was emphasized and the color-naming task may have taken on the character of preparation for a more difficult situation, that is, it became an instrumental act for the attainment of another goal. Findings from this study suggest that cognitive control deficits in some types of psychopathology are not fixed and depend crucially upon how much effort subjects are willing to mobilize in the face of stimulation that is strongly associated with feelings of failing. Psychotherapy-related asymmetries (as compared to control groups) may at least partially reflect the accumulation of experiences associating pathology-related stimuli with success that gradually outweigh their associations with failure.
When it comes to task performance, it’s important to remember that not only are task-relevant stimulus features conveyed by instruction (the “task set”), but the activation of information associated with these features (accidentally or intentionally) is also a factor in determining a task’s success. These tasks can be used in a clinical setting because they can vary widely among people and psychological conditions.