Reality Testing

Abstract

Reality testing is the capacity to distinguish internally generated mental events — memories, images, thoughts, and inner speech — from externally perceived events, and to hold one's beliefs against the evidence of the world. The term entered psychology through Sigmund Freud, who described it as an ego function separating the reality principle from wishful fantasy. Cognitive psychology reframed the idea as an information-processing problem in Marcia Johnson and Carol Raye's theory of reality monitoring and its generalization to source monitoring: the mind does not tag its contents with their origin but infers it from their qualities. This article traces that reframing, sets out the two frameworks, applies them to hallucinations through the comparator and predictive-coding accounts, reviews the prefrontal mechanisms behind the judgment, and surveys recent work treating reality testing as a metacognitive decision about internal-signal strength.

Keywords: reality testing, reality monitoring, source monitoring, hallucinations, predictive coding

Every waking moment the mind must decide which of its contents came from the world and which it produced itself, and the striking fact is how rarely it errs: a vivid memory is not mistaken for a present perception, and an imagined voice is not heard as a real one. When the discrimination fails, the consequences are severe — a hallucination is precisely an internally generated event experienced as an external perception. The concept named for this discrimination originates with Freud, who treated reality testing as the ego function that checks whether a mental image corresponds to something actually present, distinguishing the reality principle from the pleasure principle's disregard for external constraint (Freud, 1911). Cognitive psychology inherited the problem and recast it mechanistically: Johnson and Raye (1981) proposed that memories of perceived and imagined events differ systematically in their qualities, and that people decide the origin of a memory by evaluating those qualities rather than reading off a stored tag. That decision-theoretic view now organizes a literature spanning memory, clinical psychology, and cognitive neuroscience (Johnson et al., 1993; Simons et al., 2017).

What Reality Testing Is

Reality testing is the process by which a person establishes whether a mental content corresponds to the external world or has been produced by their own cognitive activity. Freud introduced the term to name an achievement of the developing ego: the infant governed by the pleasure principle hallucinates the satisfaction of a need, whereas the mature ego, governed by the reality principle, tests whether the object of a wish is actually present before acting, tolerating the delay that testing requires (Freud, 1911). In the psychoanalytic tradition the loss of this function became a defining marker of psychosis; Frosch (1964) drew the influential distinction between the neurotic, whose reality testing is intact even when reality is distorted, and the psychotic, in whom the capacity to test reality itself is impaired.

For cognitive psychology the interest is representational rather than clinical: the question is how the cognitive system, which has no privileged access to the causal history of its own states, nonetheless assigns each of them an origin. A perception and a vivid memory of the same scene are both, in the end, patterns of neural activity; nothing intrinsic labels one as outside and the other as inside. Reality testing must therefore be an inference, drawing on the differing characteristics of the two kinds of event, and this is exactly the move that turned a psychoanalytic construct into a tractable information-processing problem (Johnson & Raye, 1981; Johnson, 2006). Framed this way, reality testing is not a single faculty that is either present or absent but a decision that can be made well or poorly, that varies with the evidence available, and that can be pushed toward characteristic errors — which is why the same framework describes both ordinary misremembering and clinical hallucination.

Reality Monitoring and Source Monitoring

The cognitive reformulation begins with Johnson and Raye's (1981) theory of reality monitoring, which asks how people discriminate memories of perceived events from memories of imagined ones. Their answer is that the two classes differ on average in their qualities: memories of perceived events carry more perceptual detail (colour, sound, spatial layout) and more contextual information about time and place, whereas memories of imagined events carry more traces of the cognitive operations that produced them — the effort of construction, the reasoning that generated the image. Because these differences are statistical rather than absolute, reality monitoring is a decision under uncertainty: a memory rich in perceptual detail is attributed to perception, one bearing the marks of mental effort is attributed to imagination, and the attribution can be wrong whenever an imagined event happens to be unusually vivid or a perceived one unusually sparse.

Johnson, Hashtroudi, and Lindsay (1993) generalized this into the source-monitoring framework, which treats the reality/imagination distinction as one case of a broader problem: deciding, for any remembered content, which of several possible sources produced it — which speaker said a sentence, whether an event was witnessed or merely inferred, whether a plan was carried out or only intended. Source monitoring is likewise an attribution based on the qualitative characteristics of the memory and on the decision criteria the rememberer applies, and it comes in two modes: a fast, heuristic evaluation that operates during ordinary remembering and a slower, systematic checking that weighs the plausibility of a candidate source against other knowledge. The framework has become the standard account of a wide range of memory errors, from misattributing a suggestion to one's own experience to the everyday confusion of whether one locked the door or only meant to (Johnson, 2006). The demonstration below presents items from two sources and lets the reader vary how distinguishable the sources are, showing how source errors climb as the qualitative difference between them shrinks.

Reality Monitoring and Hallucinations

If a hallucination is an internal event experienced as an external one, then it is, in the source-monitoring vocabulary, a reality-monitoring error, and this is the core of the cognitive account of hallucinations. Bentall (1990) argued that hallucinations arise when people misattribute their own internally generated mental events — most often inner speech — to an external source, and that this bias reflects the decision criteria applied rather than any deficit of perception. The empirical prediction is direct: people prone to hallucinations should show a bias toward external attributions on source-monitoring tasks, and Bentall, Baker, and Havers (1991) found exactly that, with hallucinating patients more likely than controls to misclassify self-generated items as externally presented.

A complementary mechanism comes from Frith's (1987) comparator account, which locates the failure in the monitoring of self-generated action. On this view the motor system predicts the sensory consequences of its own commands through a forward model, and self-produced sensations are recognized as self-produced because they match the prediction; when the prediction fails to be generated or fails to be compared, self-generated events — including the covert motor act of inner speech — are not tagged as self-produced and are experienced as alien, giving rise to auditory hallucinations and to the passivity phenomena of schizophrenia in which one's own thoughts or actions feel externally controlled. The two accounts converge on a single theme, that hallucination is a breakdown of the normal attribution of internal events to the self, and later integrative models combined the source-monitoring bias, the impaired self-monitoring of action, and top-down expectation into a single cognitive model of auditory hallucinations across clinical and non-clinical populations (Aleman & Larøi, 2008; Waters et al., 2012). Table 1 sets the major cognitive and computational accounts side by side; they differ in where they locate the failure but share the claim that a hallucination is a self-generated signal wrongly attributed to the world. The demonstration then casts the judgment in signal-detection terms and lets the reader shift the decision criterion, watching a bias toward real convert internally generated items into experienced perceptions.

Table 1

Cognitive and computational accounts of hallucination as failed reality monitoring

Neural Mechanisms

The reality-monitoring decision has a reproducible neural signature in the most anterior part of the prefrontal cortex. Functional imaging localizes the discrimination of perceived from imagined information to the anterior prefrontal cortex, in and around the medial rostral region corresponding to the paracingulate sulcus, a late-developing area associated with the evaluation of internally generated cognition (Simons et al., 2017). The most striking evidence links the function to a specific anatomical feature: the paracingulate sulcus is variably present across individuals, and Garrison and colleagues (2015) found that its absence is associated with poorer reality monitoring in healthy people and, in patients with schizophrenia, with the presence of hallucinations — reduced paracingulate sulcus length predicting the symptom. The finding ties a decades-old cognitive construct to a measurable structural difference in the brain, and it does so in a way consistent with the attributional account, since the affected region is one that supports judgments about the origin of one's own mental states rather than perception itself.

Predictive Coding and Strong Priors

A more recent framework recasts reality testing within predictive processing, in which perception is not a passive reading of the senses but an inference that combines incoming sensory evidence with prior expectations, each weighted by its estimated reliability. On this account a percept is the brain's best explanation of its sensory input given what it expected, and the balance between prior and evidence is set by their relative precision (Sterzer et al., 2018). Hallucinations arise when priors are weighted too strongly: an expectation sufficiently precise can dominate weak or ambiguous sensory input and be experienced as a percept, so that the person perceives what they expect rather than what is present. Corlett and colleagues (2019) marshalled evidence for this strong priors account across sensory modalities and clinical groups, showing that people prone to hallucinations weight prior expectations more heavily in perceptual tasks and that this bias predicts the tendency to perceive signals that are not there.

The predictive-coding view reframes the reality-monitoring problem without discarding it: the question of whether a mental content is attributed to the world becomes the question of whether the balance of precision favours sensory evidence or prior expectation, and a hallucination is the endpoint at which a prior is treated as if it were sensory data. It connects the attributional account to a computational quantity — precision weighting — that can be manipulated experimentally and impaired in psychosis, and it links the clinical extremes to ordinary perception, where everyone's percepts are shaped by expectation but the weighting stays within a range that keeps prediction subordinate to evidence (Sterzer et al., 2018; Corlett et al., 2019). The demonstration lets the reader vary the relative weight of a prior expectation against noisy sensory evidence and watch the resulting percept shift from evidence-driven toward expectation-driven.

Worked Example

Consider reality monitoring cast as a signal-detection problem, the framing the second demonstration reproduces. A person is presented with a stream of items, some externally perceived and some internally generated, and must judge each as external or internal. Treat the internal-signal and external-signal strengths as two overlapping Gaussian distributions of equal variance; the separation between their means is the sensitivity d′, and the placement of the external decision boundary is the criterion c. Suppose sensitivity is fixed at d′ = 2.0, so the two means sit one standard deviation either side of zero.

With an unbiased criterion c = 0, the boundary falls midway between the distributions. The hit rate — correctly calling a true external event external — is Φ(d′/2 − c) = Φ(1.0) = 0.84, and the false-alarm rate — calling an internal event external — is Φ(−d′/2 − c) = Φ(−1.0) = 0.16. Over 100 truly internal items, that criterion produces about 100 × 0.16 = 16 misattributions of internal events to the external world. Now suppose the person adopts a liberal, hallucination-prone criterion c = −0.75, biased toward calling things real. Sensitivity is unchanged, but the false-alarm rate rises to Φ(−1.0 + 0.75) = Φ(−0.25) = 0.40, so the same 100 internal items now yield about 40 external misattributions — two and a half times as many — while the hit rate climbs only modestly to Φ(1.0 + 0.75) = Φ(1.75) = 0.96. The example makes the cognitive claim precise: a hallucination-prone shift need not degrade perceptual sensitivity at all; a change in decision criterion alone converts a substantial fraction of internally generated events into experienced external perceptions, which is the signal-detection statement of Bentall's externalizing bias (Bentall, 1990; Bentall et al., 1991).

Discussion

The history of reality testing is a case study in how a clinical construct becomes a cognitive mechanism. Freud named a function of the ego and made its loss the mark of psychosis, but the psychoanalytic version could not say how the mind tells inside from outside (Freud, 1911; Frosch, 1964). Johnson and Raye supplied the mechanism by denying that the mind stores origin at all: origin is inferred from the qualities of a memory, which makes reality testing a decision under uncertainty and, therefore, something that can be modelled, measured, and pushed into error (Johnson & Raye, 1981; Johnson et al., 1993). That single move unified normal misremembering with clinical hallucination, because both become instances of the same attribution going wrong.

The clinical and neural literatures have since converged on that attributional core from different directions. The externalizing-bias account explains hallucination as a criterion shift, the comparator account as a failure to tag self-generated action, and the predictive-coding account as an over-weighting of priors; these are not competing theories so much as descriptions of the same failure at the cognitive, computational, and mechanistic levels (Bentall, 1990; Frith, 1987; Corlett et al., 2019). The discovery that reality monitoring depends on anterior prefrontal cortex and tracks the anatomy of the paracingulate sulcus gives the whole edifice a neural anchor, connecting a construct Freud described in 1911 to a measurable feature of the individual brain (Garrison et al., 2015; Simons et al., 2017).

Current Directions

The most active current question is what, at the moment of experience, actually marks a mental content as real. Dijkstra and Fleming (2023) revived and sharpened an old idea, the perceptual-reality threshold, by showing that people judge an image to be real when its subjective signal strength exceeds a criterion that does not distinguish whether the signal came from perception or from imagination: when vivid imagery and faint perception are mixed, participants misattribute strong imagined signals to the outside world, as if reality is read off the intensity of the signal rather than its source. The result folds reality monitoring into metacognition — the monitoring of one's own cognitive states — and suggests that the classic reality/imagination distinction rests on a single graded quantity that can be experimentally displaced, a framing that connects the cognitive, clinical, and predictive-coding traditions to the study of confidence and awareness (Dijkstra & Fleming, 2023). A second, clinically driven direction is the effort to turn the strong-priors account into quantitative, individualized measures of precision weighting that could index hallucination risk before symptoms emerge, and a third is the continued mapping of reality monitoring onto its prefrontal substrate with higher-resolution imaging (Corlett et al., 2019; Simons et al., 2017).

Common Misconceptions

Reality testing is a single faculty that a person either has or lacks.

In the cognitive account it is a decision under uncertainty, made better or worse depending on the evidence and the criterion applied, not an all-or-none capacity (Johnson & Raye, 1981).

The mind stores a label recording whether each memory was perceived or imagined.

There is no stored origin tag; origin is inferred at retrieval from the qualitative characteristics of the memory, which is why the inference can be wrong (Johnson et al., 1993; Johnson, 2006).

Hallucinations reflect a failure of the sense organs.

The cognitive accounts locate hallucination in the misattribution of internally generated events to the external world, a decision-level or prediction-level failure rather than a peripheral perceptual one (Bentall, 1990; Corlett et al., 2019).

Glossary

Comparator model.

Frith's account in which self-produced sensations are recognized by matching them against a forward-model prediction of the action's consequences; a failure of the match makes self-generated events feel externally caused.

Criterion.

In signal-detection terms, the decision boundary a person applies when classifying an item; a liberal criterion for external raises both hits and false alarms.

Externalizing bias.

A tendency to attribute self-generated mental events to an external source; in Bentall's account the source-monitoring bias underlying hallucinations.

Forward model.

A prediction the motor system generates of the sensory consequences of its own commands, used to distinguish self-produced from externally produced sensation.

Hallucination.

A percept experienced in the absence of a corresponding external stimulus; in the cognitive account, an internally generated event misattributed to the external world.

Inner speech.

Covert, self-directed verbal thought; when its self-generated status is not registered it is the internal event most often misattributed to an external voice in auditory hallucination.

Paracingulate sulcus.

A variably present fold of medial anterior prefrontal cortex whose reduced length is associated with poorer reality monitoring and, in schizophrenia, with hallucinations.

Perceptual-reality threshold.

A criterion on subjective signal strength above which a mental content is judged real; it does not by itself distinguish perception from vivid imagination.

Precision weighting.

In predictive coding, the relative reliability assigned to sensory evidence versus prior expectation; over-weighting priors can make an expectation be experienced as a percept.

Predictive coding.

A framework in which perception is inference, combining sensory input with prior expectation each weighted by its precision, so that a percept is the best explanation of the input.

Reality monitoring.

The cognitive process of discriminating memories of perceived events from memories of imagined ones by evaluating their qualitative characteristics.

Reality principle.

In Freud's theory, the ego's capacity to defer gratification and act according to the constraints of the external world, in contrast to the pleasure principle.

Reality testing.

The capacity to distinguish internally generated mental events from externally perceived ones and to check belief against external evidence; impaired in psychosis.

Signal detection theory.

A framework for decisions under uncertainty that separates sensitivity (the distance between two overlapping distributions) from criterion (the response boundary); it models reality monitoring as discriminating internal from external signals.

Source monitoring.

The general process of attributing a remembered content to its origin among several possible sources, of which reality monitoring is the special case of internal versus external.

Key Researchers

Richard P. Bentall (contemporary). Clinical psychologist at the University of Sheffield whose work reframed hallucinations as a source-monitoring bias, showing that hallucinating patients over-attribute self-generated events to external sources. ORCID - Wikipedia - Wikidata

Philip R. Corlett (contemporary). Associate professor of psychiatry at Yale University whose predictive-coding, strong-priors account recasts hallucinations as overweighted prior expectations overriding sensory evidence. ORCID - Wikidata

Nadine Dijkstra (contemporary). Researcher at University College London whose work on mental imagery shows that a perceptual-reality threshold on subjective signal strength governs when an imagined signal is judged real. ORCID - Faculty page

Stephen M. Fleming (contemporary). Professor of cognitive neuroscience at University College London whose metacognition research frames reality monitoring as a metacognitive judgment about the source and strength of internal signals. Google Scholar - Faculty page

Sigmund Freud (1856-1939). Founder of psychoanalysis at the University of Vienna who introduced reality testing and the reality principle in his 1911 account of the two principles of mental functioning, framing the distinction between wish and world as an achievement of the ego. Wikipedia - Wikidata

Christopher D. Frith (contemporary). Cognitive neuroscientist at University College London whose comparator/forward-model account explains why patients with schizophrenia misattribute self-generated actions and inner speech, a mechanism of failed reality monitoring. ORCID - Wikipedia - Wikidata

Marcia K. Johnson (contemporary). Sterling Professor Emerita of Psychology at Yale University who, with Carol Raye, created the reality-monitoring framework and, with Hashtroudi and Lindsay, the source-monitoring framework — the cognitive reformulation of reality testing. Wikipedia - Google Scholar

Jon S. Simons (contemporary). Professor of cognitive neuroscience at the University of Cambridge whose work localizes reality monitoring to anterior prefrontal cortex and links paracingulate sulcus morphology to hallucinations. ORCID - Faculty page

Frequently Asked Questions

What is reality testing? Reality testing is the capacity to distinguish internally generated mental events, such as memories and images, from externally perceived events, and to check one's beliefs against the external world. The term comes from Freud, who treated it as an ego function tied to the reality principle (Freud, 1911).

How is reality testing different from reality monitoring? Reality testing is the broad clinical capacity; reality monitoring is its cognitive mechanism. Johnson and Raye proposed that people discriminate perceived from imagined memories by evaluating their qualities rather than reading a stored origin tag (Johnson & Raye, 1981).

What is source monitoring? Source monitoring is the general process of deciding where a remembered content came from among several possible sources; reality monitoring is the special case of judging internal versus external origin (Johnson et al., 1993).

How do cognitive theories explain hallucinations? They treat a hallucination as a reality-monitoring error: an internally generated event, often inner speech, misattributed to an external source because of a biased decision criterion or a failure to tag self-generated activity (Bentall, 1990; Frith, 1987).

What is the strong-priors account of hallucinations? In predictive coding, perception combines sensory evidence with prior expectation weighted by precision. When priors are weighted too heavily, an expectation can dominate weak input and be experienced as a percept, producing a hallucination (Corlett et al., 2019; Sterzer et al., 2018).

Which brain region supports reality monitoring? Anterior prefrontal cortex, in and around the paracingulate sulcus. The length of this sulcus predicts reality-monitoring accuracy in healthy people and the presence of hallucinations in schizophrenia (Garrison et al., 2015; Simons et al., 2017).

Can healthy people misattribute imagination to reality? Yes. Source errors are common in ordinary memory, and recent work shows that a sufficiently vivid imagined signal can cross the perceptual-reality threshold and be judged real even by healthy participants (Dijkstra & Fleming, 2023; Johnson, 2006).

Is loss of reality testing unique to schizophrenia? No. Impaired reality testing marks psychotic states generally, a point Frosch made in distinguishing the psychotic character from the neurotic, and auditory hallucinations occur across clinical and non-clinical populations (Frosch, 1964; Waters et al., 2012).

References