Barnes maze escape error

The question

Picture the set-up for a second. If a mouse finds the escape hole quickly, how do we know it did not just check every hole on the way? You're watching Morris Water Maze Software Guide, over at morriswatermazesoftware.com.

Meet the Barnes maze escape error

On the Barnes maze, an error is a visit to a hole that is not the escape. Primary errors are the wrong holes visited before the first visit to the escape hole. Total errors also count later visits, including revisits. Because a mouse can find the hole by working around the rim, a short latency is not enough. Error counts, especially primary errors, show whether the search was aimed or serial. A 1979 paper described the maze as a dry test of spatial memory.

How it runs

Let me walk you through it. First, define a hole visit, for example a head dip into the hole, and keep that rule for every animal. Next, count each wrong hole visited before the first visit to the escape hole as a primary error. Then, also count total errors, including revisits, and the primary latency to the first visit to the escape. Finally, in a probe, close or remove the escape box and score visits near the old hole.

Strengths and limits

So what's in it for you? Errors are discrete and easy to report. Primary errors catch serial searching that latency can hide. The same logic works with video tracking or with careful live scoring. Where does it bite? The visit rule must be fixed, or labs will not agree on the counts. Some animals explore holes out of curiosity, which adds errors that are not failed searches.

What to measure

This is where tracking software earns its keep. Primary errors: wrong holes before the first visit to the escape. Total errors: every wrong visit, including revisits. Primary latency: time to the first visit to the escape hole. Search strategy class: direct, serial or random, read from the order of hole visits. Two mice reach the escape in 40 seconds. One goes almost straight there, with two primary errors. The other works clockwise around the rim and visits ten holes first. Same latency, very different spatial accuracy.

The takeaway

Short version: total errors and its neighbours tell you how the task was done, not just how fast. The full transcript is on this page, with sources.