Elevated Plus Maze Test

Abstract

The elevated plus maze test, which MeSH classifies under maze learning, is the most widely used behavioral assay of anxiety-like behavior in rodents. Despite its filing as a maze, it measures neither route learning nor spatial memory: it is a single-trial test that exploits the conflict between a rodent's drive to explore novel space and its innate avoidance of open, elevated space. The apparatus is a plus-shaped platform raised above the floor, with two enclosed arms and two open, exposed arms; a more anxious animal keeps to the enclosed arms, while an anxiolytic drug raises the proportion of entries into, and time on, the open arms. This article covers the test's origins, how a trial is scored, its two proportion indices, the pharmacological validation that established it, and the reliability debate that shapes its use.

Keywords: elevated plus maze, anxiety-like behavior, open-arm avoidance, approach-avoidance conflict, anxiolytic

The elevated plus maze is the rodent behavioral test a pharmacologist reaches for first when asking whether a compound reduces anxiety. Its appeal is that it requires no training, no food deprivation, and no noxious stimulus: an animal placed at the center of the maze will, within a five-minute trial, distribute its exploration between the sheltered enclosed arms and the exposed open arms, and that distribution is sensitive to the same drugs that reduce anxiety in humans (Pellow et al., 1985; Walf & Frye, 2007). The test grew directly out of mid-century work on how novelty and fear jointly drive exploration, and it has become, by citation count, one of the most heavily used paradigms in all of behavioral neuroscience — which is precisely why the debate over what its scores mean, and how reproducibly they can be obtained, matters so much (Montgomery, 1955; Carobrez & Bertoglio, 2005).

What the Elevated Plus Maze Test Is

The elevated plus maze test is a measure of anxiety-like behavior in which a rodent is placed on a raised, plus-shaped apparatus and its avoidance of the open, unprotected arms is scored over a brief trial. The maze has four arms of equal length arranged in a cross: two opposite arms are enclosed by high side walls, and the other two are open, bordered at most by a slight lip. The whole structure stands well above the floor — roughly half a meter for rats — so that the open arms combine three aversive features a rodent innately avoids: openness, height, and, under typical lighting, brightness. A single animal is placed at the central square, usually facing an open arm, and left to explore freely, most often for five minutes (Walf & Frye, 2007).

What makes the arrangement informative is that it pits two innate tendencies against each other. A rodent in a novel environment is driven to explore it, and the open arms are the most novel part of the maze; but the same animal is driven to avoid exposed, elevated space and to keep to walls and edges — a wall-hugging tendency called thigmotaxis — which the enclosed arms let it do. The behavior that results — how much of its exploration the animal is willing to direct toward the open arms — is an expression of this approach-avoidance conflict, and it is read as an index of the animal's anxiety state (Montgomery, 1955; Rodgers & Dalvi, 1997). Crucially, no learning is required and none is measured: the test captures a spontaneous behavioral choice on first exposure, not the acquisition of a route, which is why it sits uneasily under its MeSH parent, maze learning. It is filed there because of its shape and lineage, not because it assays the same process.

The test is fundamentally an emotional and motivational assay rather than a cognitive one. Its validity rests on the claim that open-arm avoidance reflects a state analogous to human anxiety, and the strongest support for that claim is pharmacological: drugs that relieve anxiety in people selectively increase open-arm exploration, and drugs that provoke anxiety decrease it (Pellow et al., 1985).

How the Test Works

The paradigm descends from a specific insight about exploration. Montgomery observed that novel stimulation evokes both an approach tendency, expressed as exploration, and a competing fear tendency, expressed as avoidance, and that the balance between them could be measured by giving an animal a choice between more and less exposed alleys (Montgomery, 1955). Handley and Mithani built the modern elevated plus maze on that foundation, using the animal's differential avoidance of open versus enclosed arms as a model of fear-motivated behavior and showing it responded to noradrenergic drugs (Handley & Mithani, 1984). The decisive step was taken by Pellow, Chopin, File, and Briley, who subjected the rat plus maze to a full pharmacological and behavioral validation and established the scoring conventions still in use (Pellow et al., 1985). Lister then adapted the apparatus and dimensions for the mouse, extending the test to the species that dominates modern genetic research (Lister, 1987).

A trial is deliberately simple. The maze is cleaned between animals to remove odor cues, the subject is placed at the center, and its movements are recorded — historically by a trained observer, now almost always by video tracking software. An arm entry is conventionally scored when the animal crosses onto an arm with all four paws. The session is short because the measure depends on novelty: the aversive quality of the open arms erodes with experience, so the informative behavior occurs in the first few minutes of the first-ever exposure (Walf & Frye, 2007; Kraeuter et al., 2019). This single-trial character is not a convenience but a constraint built into what the test measures.

That constraint has a striking consequence known as one-trial tolerance: an animal tested a second time no longer shows an anxiolytic response to benzodiazepines, even though its baseline open-arm avoidance may look similar. Prior experience of the maze changes the nature of the task — from a response to novelty into something closer to a learned avoidance — so the drug-sensitive process is present only on first exposure (Rodgers & Dalvi, 1997). For this reason the elevated plus maze is treated as a strictly between-subjects, single-use test for each animal.

The Core Measures

The test's power rests on expressing behavior as proportions rather than raw counts. The two primary measures are the percentage of open-arm entries (%OAE), the number of entries into open arms divided by the total number of arm entries, and the percentage of open-arm time (%OAT), the time spent on the open arms divided by the time spent on any arm. Both rise when anxiety is reduced and fall when it is increased (Pellow et al., 1985; Walf & Frye, 2007). Proportions are used because they are far less contaminated by overall activity than raw counts are: an animal that simply moves more will make more entries of every kind, inflating raw open-arm entries without any change in its relative preference.

The critical control measure is the number of closed-arm entries, which Pellow and colleagues identified as the index of general locomotor activity that is most independent of the anxiety measures (Pellow et al., 1985). The logic of interpretation follows directly: a genuine anxiolytic raises %OAE and %OAT while leaving closed-arm entries roughly unchanged, whereas a drug that merely stimulates locomotion raises closed-arm entries and total activity without selectively increasing the proportion of behavior directed at the open arms. A compound that increased open-arm entries only in step with a general rise in all entries would be a stimulant masquerading as an anxiolytic, and the closed-arm-entry control is what unmasks it (Rodgers & Dalvi, 1997).

Table 1

Reading the Plus-Maze Measures Against the Locomotor Control

Note. The proportion measures (%OAE, %OAT) index anxiety-like behavior, while closed-arm entries index general locomotion. A selective rise in the proportions with stable closed-arm entries is the anxiolytic signature; reading the two together is what separates a genuine anxiety effect from a change in activity.

Beyond these classical indices, a second generation of ethological measures was added to sharpen the test. Rather than counting only entries and time, researchers score risk-assessment behaviors — stretched-attend postures, head-dipping over the edge of an open arm, and flat-back approach — which track the anxiety state more finely than the spatiotemporal measures alone and help distinguish a true change in emotionality from a change in motor output (Rodgers & Dalvi, 1997; Carobrez & Bertoglio, 2005).

Validity and Neural Basis

The elevated plus maze earned its standing through predictive validity: its scores respond to drugs in the direction their clinical effects predict. Clinically effective anxiolytics such as the benzodiazepines selectively increase open-arm exploration, while anxiogenic compounds, and stimuli that heighten arousal, decrease it — a bidirectional pharmacological sensitivity established in the original validation and confirmed across hundreds of subsequent studies (Pellow et al., 1985; Pellow & File, 1986). This drug sensitivity, together with the test's simplicity and speed, is why it became the field's default anxiety screen.

The behavior draws on a defensive system centered on the amygdala and its connections to the hippocampus and prefrontal cortex, the same circuitry that governs innate fear and avoidance more generally. Open-arm avoidance is understood as an expression of this circuit's response to a naturally threatening situation, and manipulations of these structures and of the neurotransmitter systems that modulate them — GABAergic, serotonergic, and noradrenergic — shift open-arm behavior in predictable ways (Handley & Mithani, 1984; Carobrez & Bertoglio, 2005).

Validity, however, is not the same as reliability, and the test's greatest vulnerability lies in the second. Hogg's early review documented how strongly plus-maze scores depend on procedural details — illumination, whether the open arms have a raised ledge, prior handling, the time of day, even the strain of animal — and how these factors muddy comparison across laboratories (Hogg, 1996). The problem is not that the test is invalid but that it is delicate: the same nominal protocol can yield different baselines in different hands, so that between-laboratory disagreement can reflect the apparatus and setting as much as the biology under study. The sharpest demonstration came from Crabbe and colleagues, who ran identical mouse strains on the same behavioral tests — including a plus-maze measure — in three laboratories that had rigorously equated apparatus, protocol, and handling, and still found systematic between-laboratory differences and lab-by-genotype interactions, a result that made the reproducibility problem impossible to dismiss as mere sloppiness (Crabbe et al., 1999).

Worked Example

Consider how the proportion measures separate anxiety from activity, using two five-minute (300 s) trials. A vehicle-treated control rat makes 3 open-arm entries and 9 closed-arm entries, spending 45 s on the open arms and 180 s in the closed arms. Its total arm entries are 3 + 9 = 12, so its open-arm-entry percentage is %OAE = 3/12 = 25.0%. Its total arm time is 45 + 180 = 225 s, so its open-arm-time percentage is %OAT = 45/225 = 20.0%. These low values are the signature of a relatively anxious animal that keeps to shelter.

Now give a second rat an effective anxiolytic. It makes 8 open-arm entries and 12 closed-arm entries, spending 108 s on the open arms and 132 s in the closed arms. Its %OAE = 8/(8+12) = 8/20 = 40.0% and its %OAT = 108/(108+132) = 108/240 = 45.0% — both sharply higher, the expected anxiolytic profile. The activity control confirms the interpretation: closed-arm entries rose only modestly, from 9 to 12, so the shift toward the open arms is not merely a byproduct of moving more.

The contrast with a pure locomotor stimulant makes the point precisely. Suppose a stimulant simply doubles every kind of movement in the control animal: open-arm entries 3 → 6 and closed-arm entries 9 → 18, for 24 total entries. The raw open-arm-entry count has doubled, which a naive reading might mistake for reduced anxiety — yet %OAE = 6/24 = 25.0%, exactly the control value, because the proportion is invariant to a uniform scaling of activity. Meanwhile closed-arm entries have doubled to 18, flagging the locomotor increase directly. This is why the proportion measures, read against the closed-arm-entry control, are the interpretable output of the test, and raw open-arm counts alone are not (Pellow et al., 1985; Rodgers & Dalvi, 1997).

Discussion

The elevated plus maze occupies a peculiar and instructive position in behavioral science: it is at once one of the most useful and one of the most criticized tests in the field. Its usefulness is real and hard won. A test that needs no training, runs in five minutes, requires no aversive reinforcement, and responds bidirectionally to anxiolytic and anxiogenic drugs is a genuinely powerful screening tool, and the pharmacological validation Pellow and colleagues supplied gave it a firm evidential basis that many behavioral models lack (Pellow et al., 1985; Carobrez & Bertoglio, 2005). Its speed and simplicity are not superficial virtues; they are what let a single measure be applied across pharmacology, genetics, and lesion studies with a common currency.

The criticisms are equally real and cut in two directions. The first is construct-level: open-arm avoidance is a single behavioral output standing in for the complex human construct of anxiety, and there is no guarantee the mapping is clean, particularly given one-trial tolerance and the test's insensitivity to some clinically effective drugs (Rodgers & Dalvi, 1997). The second is methodological: the delicacy Hogg documented means that a plus-maze score is a joint function of the animal's biology and the exact conditions of testing, so that reproducibility across laboratories cannot be assumed (Hogg, 1996). Neither criticism retires the test; both sharpen the reading of it. The maze remains indispensable precisely because its limits are so well charted — a rare case where a tool's weaknesses are documented as thoroughly as its strengths.

Current Directions

Contemporary work on the elevated plus maze is dominated by the reproducibility problem the test has come to exemplify. Because plus-maze data are so sensitive to procedural and environmental variables, recent effort has gone into detailed standardized protocols intended to make results comparable across laboratories, specifying apparatus dimensions, lighting, handling, and scoring criteria in a way the original reports left implicit (Kraeuter et al., 2019). A parallel line asks whether alternative apparatus geometries are inherently more stable: the elevated zero maze, which removes the central square and its ambiguous entries, has been reported to yield more consistent behavior across repeated trials than the plus maze in some comparisons, though at the cost of losing the closed-arm-entry activity index (Tucker & McCabe, 2017).

A more pointed critique reframes the reproducibility debate itself. Rather than treating cross-laboratory disagreement as noise to be engineered away, some argue that the variability is telling us the test measures the response to a particular apparatus in a particular setting as much as it measures a stable anxiety trait — so that the pharmacology being characterized is, in part, the pharmacology of the plus maze rather than of anxiety in general (Bespalov & Steckler, 2021). On this view the productive response is not only tighter standardization but converging evidence across multiple, mechanistically distinct assays, with the plus maze as one contributor rather than a sole arbiter. The debate keeps a forty-year-old test at the center of current methodological reform in behavioral pharmacology.

Common Misconceptions

The elevated plus maze measures maze learning or spatial memory.

It does not. Despite being classified under maze learning in MeSH because of its shape, the test involves no learning of a route and no spatial memory. It is a single-trial measure of anxiety-like behavior that exploits a rodent's innate avoidance of open, elevated space (Pellow et al., 1985; Walf & Frye, 2007).

More open-arm entries always mean less anxiety.

Only if activity is held constant. A pure locomotor stimulant raises the raw count of every kind of entry, including open-arm entries, without reducing anxiety. This is why the interpretable measures are proportions (%OAE and %OAT), read against closed-arm entries as an index of general activity (Pellow et al., 1985; Rodgers & Dalvi, 1997).

An animal can be retested on the maze to replicate its score.

The test is strictly single-use. Prior exposure produces one-trial tolerance: on a second trial the anxiolytic response to benzodiazepines disappears, because experience changes the maze from a novel space into a familiar one. Each animal contributes one trial in a between-subjects design (Rodgers & Dalvi, 1997).

Glossary

Amygdala.

A medial temporal lobe structure central to the defensive circuit that generates innate fear and open-arm avoidance in the plus maze.

Anxiety-like behavior.

A rodent behavioral state, analogous to human anxiety, inferred from avoidance of aversive situations; in the plus maze it is indexed by avoidance of the open arms.

Anxiolytic.

A drug that reduces anxiety; in the plus maze an effective anxiolytic selectively increases open-arm entries and time without raising general activity.

Approach-avoidance conflict.

The competition between the drive to explore novelty and the drive to avoid a threatening situation; the plus maze reads anxiety from how this conflict is resolved.

Benzodiazepine.

A class of clinically effective anxiolytic drug that acts on GABAergic transmission; its selective enhancement of open-arm exploration was central to validating the plus maze.

Closed-arm entries.

The number of entries into the enclosed arms, used as the index of general locomotor activity that is most independent of the anxiety measures.

Elevated zero maze.

A variant apparatus shaped as a raised annulus with alternating open and closed quadrants; it removes the ambiguous central square and, in some reports, yields more consistent behavior.

Ethological measures.

Fine-grained behaviors such as head-dipping, stretched-attend posture, and risk assessment that track anxiety more sensitively than entries and time alone.

One-trial tolerance.

The loss of the anxiolytic drug response on a second exposure to the maze; prior experience changes the task, making the test valid only on first exposure.

Open-arm avoidance.

The tendency to spend less time and make fewer entries on the exposed open arms; greater avoidance indicates a more anxious state.

Percentage of open-arm entries (%OAE).

Open-arm entries divided by total arm entries; a primary anxiety index that is invariant to a uniform change in overall activity.

Percentage of open-arm time (%OAT).

Time on the open arms divided by total arm time; the second primary index, rising with reduced anxiety and falling with increased anxiety.

Predictive validity.

The property that a test's scores respond to manipulations in the direction their real-world effects predict; the plus maze has it because anxiolytics and anxiogenics move open-arm behavior as expected.

Thigmotaxis.

The tendency to stay in contact with walls and edges; in the plus maze it is expressed as the preference for the enclosed arms and their walls.

Key Researchers

Robert John Rodgers. University of Leeds (emeritus); his ethological reanalysis of the plus maze introduced risk-assessment scoring and documented one-trial tolerance, reshaping how the test is run and interpreted. Faculty page - Google Scholar

Antonio de Pádua Carobrez. Universidade Federal de Santa Catarina; his twenty-year reappraisal of the plus maze consolidated the ethological and temporal analysis of anxiety-like behavior and remains a standard reference for the test's methodology. Faculty page

Leandro José Bertoglio. Universidade Federal de Santa Catarina; his work on the temporal structure of plus-maze behavior and its neural substrates advanced the analysis of how anxiety-like responses unfold within a trial. Google Scholar

Cheryl A. Frye. University at Albany, SUNY; her standardized Nature Protocols description of the elevated plus maze assay is among the most widely followed procedural references for the test. Faculty page - Google Scholar

Frequently Asked Questions

What is the elevated plus maze test?
The elevated plus maze is a behavioral test of anxiety-like behavior in rodents. The animal is placed on a raised, plus-shaped platform with two enclosed and two open arms, and its avoidance of the exposed open arms over a five-minute trial indexes its anxiety state (Pellow et al., 1985; Walf & Frye, 2007).

Does the elevated plus maze measure learning?
No. Although MeSH files it under maze learning because of its shape, the test involves no route learning and no spatial memory. It is a single-trial assay of a spontaneous behavioral choice driven by the conflict between exploring novelty and avoiding open space (Montgomery, 1955; Rodgers & Dalvi, 1997).

What are the main measures scored on the maze?
The two primary measures are the percentage of open-arm entries (%OAE) and the percentage of open-arm time (%OAT), both of which rise when anxiety is reduced. The number of closed-arm entries is scored separately as an index of general locomotor activity (Pellow et al., 1985).

Why are proportions used instead of raw counts?
Because raw counts are contaminated by overall activity: an animal that moves more makes more entries of every kind. Expressing open-arm behavior as a proportion of total behavior removes that confound, so %OAE and %OAT reflect relative preference rather than sheer activity (Rodgers & Dalvi, 1997).

How was the test validated?
Pellow, Chopin, File, and Briley validated the rat plus maze pharmacologically in 1985, showing that clinically effective anxiolytics increase open-arm exploration while anxiogenic compounds decrease it. This bidirectional drug sensitivity is the test's core evidence for predictive validity (Pellow et al., 1985; Pellow & File, 1986).

What is one-trial tolerance?
One-trial tolerance is the finding that an animal retested on the maze no longer shows an anxiolytic response to benzodiazepines. Prior experience changes the maze from a novel space into a familiar one, so the drug-sensitive process exists only on the first exposure, making the test single-use (Rodgers & Dalvi, 1997).

Why are the test's results sometimes hard to reproduce?
Plus-maze scores are highly sensitive to procedural and environmental details — lighting, handling, apparatus design, strain, and time of day — so the same nominal protocol can give different baselines in different laboratories. Detailed standardized protocols aim to reduce this variability (Hogg, 1996; Kraeuter et al., 2019).

How does the elevated plus maze relate to the amygdala?
Open-arm avoidance depends on a defensive circuit centered on the amygdala and its connections to the hippocampus and prefrontal cortex. Manipulating these structures, or the GABAergic, serotonergic, and noradrenergic systems that modulate them, shifts open-arm behavior in predictable ways (Handley & Mithani, 1984; Carobrez & Bertoglio, 2005).

References