ConductScienceConductScience
ToolsConductScience tool
% CorrectFree in-browser calculator

T-Maze Percent Correct Calculator.

Enter per-trial choices and rewarded side schedule. Get % correct per block, learning curves, trials to criterion, and reversal learning analysis.

Run calculator Request a Quote
PrivateData stays in your browser
LiveNo sign-up required
Open methodFormula shown on this page
CitableMethods and citation included

Calculator

Results update in place

Try it out

Load example T-Maze data to see the full workflow

Configuration

Criterion = 8/10 correct

Add Trial

When to use

  • Compute % correct per block from per-trial left/right choice data in a T-maze position discrimination task
  • Track learning curves across blocks during acquisition and reversal phases
  • Determine trials to criterion for each animal and compare between groups
  • Analyze reversal learning by switching the rewarded side after criterion is reached
  • Compute mean choice latency per block as a secondary decision-making measure
  • Compare acquisition and reversal performance between treatment groups (e.g., WT vs. KO)
  • Export per-animal and group summary results to CSV for further statistical analysis

Do not use for

  • Spontaneous alternation (Y-maze or continuous T-maze without discrete trials) — use a spontaneous alternation calculator instead
  • Radial arm maze tasks with more than 2 choice arms — use a radial maze error calculator
  • Water maze or Barnes maze spatial reference memory — different paradigm, different metrics

Control for odor cues between trials

Rodents have exceptional olfactory ability. If the maze is not cleaned between trials, animals may use odor trails from previous runs rather than learning the spatial rule. Wipe the maze with 70% ethanol or rotate maze arms between trials. Failure to control odor cues can produce artificially high percent correct that does not reflect spatial learning.

Use pseudo-random reward schedules to prevent side biases

If the rewarded side is always left, some animals develop a position habit rather than learning the discrimination. Use a pseudo-random sequence (e.g., Gellermann series) where each block has equal left and right rewarded trials with no more than 3 consecutive same-side rewards. This calculator allows you to set the rewarded side per trial or per block.

Report block size and criterion explicitly

Trials to criterion depends entirely on block size and criterion threshold. A criterion of 8/10 in blocks of 10 is standard, but some labs use 9/10 or require 2 consecutive blocks at criterion. Always report these parameters. This calculator lets you configure both.

Distinguish perseverative from random errors in reversal

During reversal, early errors are typically perseverative (the animal continues choosing the previously correct side). Later errors may be random. Perseverative errors in the first reversal block are a specific index of cognitive inflexibility. This calculator tracks per-block accuracy so you can see whether the initial reversal block shows below-chance performance (perseveration) before recovery.

Ensure food restriction is consistent across groups

T-maze performance depends on motivation. Animals must be food-restricted (typically to 85-90% of free-feeding weight) and maintained at stable weights throughout testing. Differential food restriction between groups can confound choice accuracy and latency differences.

1

Method

Each trial is scored as correct (choice matches rewarded side) or incorrect. Percent correct per block equals correct trials in the block divided by block size, multiplied by 100. Trials to criterion is the cumulative trial count at the end of the first block where percent correct meets or exceeds the criterion threshold. For reversal analysis, the rewarded side flips after criterion and a new learning curve begins. Mean choice latency per block is the arithmetic mean of per-trial latencies within each block. Group statistics use sample standard deviation (n-1 denominator) for SEM computation. All computation is client-side — no data leaves your browser.

2

Published

Published 2026-04-05. Calculations support planning and documentation; verify procurement decisions against manufacturer specifications or institutional SOPs.

3

How to cite

How to Cite

ConductScience T-Maze Percent Correct Calculator. ConductScience, Inc. 2026. Available at: https://conductscience.com/tools/t-maze-percent-correct-calculator

This tool performs mathematical calculations on user-provided data. It does not replace scientific judgment regarding experimental design, exclusion criteria, or statistical analysis.

What Is the T-Maze?

The T-maze is one of the oldest and most straightforward maze paradigms in behavioral neuroscience, first used systematically in the early 20th century. Its simplicity — a start arm leading to a binary left/right choice — makes it ideal for studying position discrimination, spatial reference memory, working memory (in delayed alternation variants), and reward-guided decision-making. The apparatus can be constructed with opaque walls and removable barriers or guillotine doors to control access. In the rewarded position discrimination task, one goal arm consistently contains a food reward, and the animal must learn to choose the correct side across multiple trials. The T-maze is widely used in mice and rats, with well-established normative data for various strains. Its simplicity also makes it suitable for high-throughput phenotyping, pharmacological screening, and lesion studies, particularly for hippocampal and prefrontal cortex function.

The Discrete-Trial Procedure

In the discrete-trial T-maze task, each trial is a single decision event. The animal starts in the start arm behind a closed door. When the door opens, the animal runs forward and enters either the left or right goal arm. If the animal chooses the rewarded arm, it consumes the food reward (typically a small pellet or sweetened condensed milk well). If it chooses the unrewarded arm, it encounters an empty well and is removed after a brief confinement period. The animal is then returned to a holding cage for an inter-trial interval (ITI), typically 15-30 seconds. Trials are grouped into blocks, commonly 10 trials per block, and sessions typically consist of 1-4 blocks (10-40 trials) depending on the protocol. Key experimental controls include: (1) pseudo-random reward schedules to prevent side biases (e.g., no more than 3 consecutive rewarded-left trials), (2) cleaning the maze between trials to eliminate odor cues, and (3) consistent handling procedures. The primary outcome is percent correct per block. Secondary outcomes include choice latency (time from door opening to arm entry) and errors of commission vs. omission.

Reversal Learning in the T-Maze

Reversal learning is a critical extension of the basic T-maze position discrimination task. Once the animal reaches a predetermined learning criterion (e.g., 80% correct for one block) on the initial acquisition, the contingencies are reversed: the previously unrewarded arm now contains the reward, and the previously correct arm is now empty. The animal must inhibit the previously learned response and acquire the new rule. This dissociation is experimentally powerful because initial acquisition depends primarily on the hippocampus and dorsal striatum, while reversal learning additionally recruits the orbitofrontal cortex (OFC) and ventromedial prefrontal cortex (vmPFC). Animals with OFC lesions or serotonergic depletion typically show normal acquisition but impaired reversal — they perseverate on the previously correct side for more trials. The key metrics for reversal learning are: (1) trials to criterion on reversal (compared to acquisition), (2) the number of perseverative errors (choosing the previously correct, now incorrect, side) in the first reversal block, and (3) the shape of the reversal learning curve, which typically shows an initial dip below chance (perseveration) followed by recovery. Comparing acquisition trials-to-criterion with reversal trials-to-criterion quantifies cognitive flexibility.

Frequently asked

What is a T-maze and how is it used in behavioral research?
A T-maze is a simple maze shaped like the letter T, consisting of a start arm and two goal arms (left and right). The animal is placed at the base of the start arm and must choose between the left or right goal arm on each trial. T-mazes are used to study spatial learning, working memory, and decision-making in rodents. In a discrete-trial rewarded alternation or position discrimination task, one goal arm is baited with a food reward, and the animal must learn which side is correct. The primary measure is percent correct — the proportion of trials on which the animal chooses the rewarded side.
What is the discrete-trial procedure in a T-maze task?
In the discrete-trial procedure, each trial is a single run from the start arm to one goal arm. The animal is placed at the start, a door is opened, and the animal runs and makes a left or right choice. After the choice, the animal is returned to a holding cage for an inter-trial interval (typically 15-30 seconds) before the next trial. Trials are grouped into blocks (commonly 10 trials per block), and percent correct is computed per block. The criterion for learning is typically defined as a threshold percent correct (e.g., 80% or 8/10) sustained for one or more consecutive blocks.
What is reversal learning and why is it important?
Reversal learning is a phase that begins after the animal reaches criterion on the initial discrimination. The rewarded side is switched (e.g., from left to right), and the animal must learn the new contingency. Reversal learning measures cognitive flexibility — the ability to suppress a previously learned response and acquire a new one. It engages prefrontal cortex and orbitofrontal cortex. Animals with lesions or pharmacological manipulations affecting these regions often show impaired reversal learning even when initial acquisition is intact. Comparing trials to criterion in acquisition vs. reversal is a standard analysis.
How is percent correct calculated per block?
Percent correct per block equals the number of correct trials divided by the total number of trials in that block, multiplied by 100. A trial is correct when the animal chooses the same side as the rewarded side for that trial. For example, if the rewarded side is left and the animal chooses left, that trial is correct. With a block size of 10, if an animal makes 7 correct choices, percent correct for that block is 70%. Chance performance in a two-choice task is 50%, so the chance line on learning curves is drawn at 50%.
What is a reasonable criterion for learning in the T-maze?
The most common criterion is 80% correct (8 out of 10 trials) for one or two consecutive blocks. Some studies use stricter criteria such as 90% (9/10) or require the criterion to be met across two consecutive blocks to reduce false positives from lucky guessing. The criterion should be specified in your methods section. This calculator reports trials to criterion as the trial number at the end of the first block that meets or exceeds the specified criterion percentage.
How should I interpret choice latency in the T-maze?
Choice latency is the time from trial start (door opening or animal release) to the moment the animal enters a goal arm. Short latencies with high accuracy suggest confident, well-learned responses. Long latencies may indicate deliberation (vicarious trial and error, or VTE), especially during early learning or at the start of reversal. Increasing latencies during reversal often reflect response conflict between the old and new rules. Mean latency per block is reported alongside percent correct to give a more complete picture of the animal's decision process. Latency entry is optional in this calculator.
Can I analyze multiple animals and groups simultaneously?
Yes. Enter trial data for each animal with an Animal ID and Group label (e.g., WT or KO). The calculator computes per-block percent correct, trials to criterion, and mean latency for each animal individually, then aggregates group-level statistics (mean plus or minus SEM). The learning curve chart overlays group means across blocks, and the bar chart compares trials to criterion between groups. This allows direct comparison of acquisition and reversal performance across experimental conditions.
325
Free tools
1,200+
Institutions
100%
Client-side
0
Uploads required

Next steps

Equipment

T-Maze

T-maze apparatus for discrete-trial choice behavior and spatial memory testing

Browse Equipment

Automate with ConductVision

ConductVision tracks arm entries and choice latency automatically from overhead video

Browse Equipment

All behavior tests by paradigm

Browse every rodent behavior calculator, analyzer, and scorer organized by anxiety, depression, motor, memory, social, and pain paradigms.

Browse Equipment

Run a behavior core

Battery sequencer, acclimation tracker, IRR calculator, ethogram builder, and SOP generator.

Browse