Radial Arm Maze Reference Memory
Overview
The radial arm maze reference memory task simultaneously assesses both working memory and long-term reference memory by baiting a fixed subset of arms (typically 4 of 8) consistently across all sessions while leaving the remaining arms permanently unbaited. Over multiple training sessions, the animal must learn two concurrent rules: which arms are consistently baited (reference memory) and which of those baited arms have been visited within the current session (working memory). Entries into never-baited arms constitute reference memory errors, while re-entries into already-visited baited arms constitute working memory errors, enabling independent measurement of these two memory systems within the same task.
This dual-error classification is the signature feature of the protocol and allows pharmacological or genetic manipulations to be characterized by their selective effects on one memory type versus the other. Cholinergic disruption preferentially increases working memory errors, while hippocampal lesions elevate both error types. Reference memory performance improves across days as the animal learns the fixed arm configuration, and the acquisition curve slope provides an index of spatial learning rate. Working memory errors remain relatively stable once training is complete because the within-session demands do not change with further experience.
ConductMaze implements the dual-memory radial arm maze protocol with configurable arm-baiting assignments, automated reward detection, and real-time error classification. The software distinguishes working memory from reference memory errors on a choice-by-choice basis, generates separate learning curves for each error type across training days, and computes error ratios that reveal the relative integrity of each memory system. Arm-baiting configurations are locked across sessions to ensure consistent reference memory demands, and the software flags animals that develop non-spatial strategies such as serial arm entry patterns.
Trial Flow
Selective Baiting
A fixed subset of arms (e.g., 4 of 8) is baited; the same arms are baited every session.
Central Release
Animal is placed in the central hub with access to all 8 arms.
Choice Recording
Each arm entry is recorded and classified as baited-first-visit, baited-revisit, or never-baited.
Error Classification
Entry into a never-baited arm = reference memory error; revisit to a depleted baited arm = working memory error.
Reward Collection
Pellet sensors confirm reward consumption on first visits to baited arms.
Session Completion
Session ends when all baited arms are visited or maximum choices are reached.
Metrics Computation
Working memory errors, reference memory errors, and learning curve data are computed.
Session End
Animal is removed; data are exported and added to the multi-day learning curve.
Parameters
| Parameter | Type | Default | Description |
|---|---|---|---|
| Number of Arms | integer | 8 | Total number of arms in the radial maze. |
| Number Baited | integer | 4 | Number of arms that contain food reward (consistent across sessions). |
| Baited Arms | enum | counterbalanced | Which specific arms are baited: fixed across subjects, counterbalanced, or randomized per subject. |
| Maximum Choices | integer | 12 | Maximum arm entries per session before termination. |
| Session Timeout | duration | 10min | Maximum session duration. |
| Reward Type | enum | sucrose-pellet-45mg | Food reward placed in baited arms. |
| Training Days | integer | 15 | Total number of training sessions across which learning curves are generated. |
| Food Restriction Target | percentage | 85% | Body weight target as percentage of free-feeding weight. |
| Arm Length | distance | 45cm | Length of each arm from central hub to reward location. |
| Criterion | integer | 1 | Maximum total errors on two consecutive sessions to reach learning criterion. |
Metrics
| Metric | Unit | Description |
|---|---|---|
| Reference Memory Errors | count | Number of entries into never-baited arms per session. |
| Working Memory Errors | count | Number of re-entries into already-visited baited arms per session. |
| Total Errors | count | Sum of reference and working memory errors per session. |
| Days to Criterion | count | Number of training days to reach the defined learning criterion. |
| Correct in First 4 | count | Number of baited arms visited within the first 4 choices (max = 4 if optimal). |
| RM/WM Error Ratio | ratio | Ratio of reference to working memory errors, characterizing the dominant deficit type. |
| Mean Choice Latency | s | Average time between consecutive arm entries. |
| Time to Complete | s | Total time to collect all rewards or reach session termination. |
Sample Data
| Subject | Group | RM Errors | WM Errors | Days to Criterion | Correct in First 4 |
|---|
Representative data for illustration purposes. Actual values will vary by species, strain, and experimental conditions.
Applications
- 1Dual Memory System Assessment — Simultaneously measuring working and reference memory to characterize the specific memory system affected by a manipulation.
- 2Cholinergic Selectivity — Demonstrating selective working memory impairment by muscarinic antagonists with relatively spared reference memory.
- 3Hippocampal vs Striatal Function — Dissociating hippocampal contributions (both error types elevated) from striatal contributions (reference memory selectively impaired).
- 4Chronic Drug Studies — Tracking the development and persistence of memory impairments across weeks of daily testing with separate learning curves for each error type.
- 5Cognitive Aging — Mapping the differential vulnerability of working versus reference memory to normal aging processes.
Related Protocols
Compatible Products
Ready to Automate Your Behavioral Protocols?
Contact us for a demo and pricing information.
