Knowledge Base

Problem

Users running T-maze, radial arm maze, or similar choice-based experiments want to define 'errors' (e.g., entering the wrong arm or moving away from the goal) and quantify them automatically in EthoVision XT.

Details

EthoVision XT does not have a built-in 'error' parameter, but errors can be defined using zone-based approaches (entry into incorrect zones) or using conditional/event variables. For maze experiments where the animal must choose a correct goal arm, wrong choices can be captured as entries into non-target zones. More nuanced error definitions (e.g., continuous movement away from goal) may require combining zone entries with directional or distance parameters. Applies to T-maze, radial arm maze, and other choice-based paradigms across EthoVision XT versions.

Solution

**Approach 1 – Zone-based error counting (recommended for discrete wrong-arm entries):**
1. Define the correct goal arm/zone and each incorrect arm/zone in your arena setup.
2. Use the 'In Zone' parameter for each incorrect zone to detect when the animal enters a wrong arm.
3. Create a calculated variable or use the 'Number of Entries' parameter for incorrect zones — each entry represents one error.
4. In your Analysis Profile, include zone entry counts for non-target zones as your error measure.

**Approach 2 – Conditional variable for complex error definitions:**
1. If an error involves continuous movement away from the goal (not just discrete zone entries), create a Boolean variable that is TRUE when the animal is in any non-target zone.
2. Use a conditional variable to accumulate time or count events when this condition is met.
3. Combine with a minimum duration threshold if brief passes through non-target areas should not count as errors.

**General tips:**
- Label zones clearly (e.g., 'Goal Arm', 'Error Arm 1', 'Error Arm 2') to make analysis profiles intuitive.
- Validate your error detection by reviewing a subset of trials manually and comparing automated counts to observed choices.
- Export error counts along with other performance measures (latency, distance) for statistical analysis.