ANY-maze Help > The ANY-maze reference > The Protocol page > The elements of a protocol > Behaviour > Freezing detection

Freezing detection

Introduction

ANY-maze can automatically detect periods when the animal is freezing, and include analysis of them in the test results.

You can specify whether you want it to do this using the options in the Freezing detection element of the protocol.

Determining when an animal is freezing

Freezing detection in ANY-maze is performed on an apparatus-wide basis. Essentially, ANY-maze looks for movement anywhere in the apparatus, and when it can't find any it considers the animal to be freezing. Of course, the system takes into account that the video image will include some noise (the individual pixels tend to flicker a little) and that even when freezing, an animal will still breathe and therefore it will move very slightly.

The mechanics of freezing detection require that the system finds a sequence of images in which there is no movement before it will actually score the animal as 'freezing'. This means that the moment when it concludes that the animal is freezing will always be a short time after the freezing actually began. ANY-maze automatically adjusts for this delay (which varies, but is typically around 200ms) such that the start of a freezing bout will be scored at the true start, and not at the 'detection' start.

Choosing the freezing detection system to use

We are continuously working on the freezing detection system in ANY-maze to make it both simpler to use and more accurate. When you create a new experiment, ANY-maze will automatically select the most recent 'freezing engine' and in most cases this is what we recommend you use.

However, you may wish to use a different freezing engine if you want your experiment to be performed in exactly the same way as a previous experiment.

For example, if you performed an experiment in July 2021 using ANY-maze version 7.00, then it would have used the V7 freezing engine. If it is now October 2022 and you are using ANY-maze version 7.20, then it will automatically choose the V7.2 freezing engine for use in a new experiment. But this means that the freezing results in this new experiment will be scored a little differently to those from the earlier experiment. To avoid this you could change the new experiment to use the V7 freezing engine, making the freezing detection in the two experiments identical.

IMPORTANT: To ensure that all the tests in an experiment use the same freezing detection system, you can't change the system once you have begun testing.

Differences between the ANY-maze freezing engines:

Setting the minimum freeze duration

ANY-maze will detect even very short periods of freezing, but you will probably not consider these to be 'freezing' in the behavioural sense, it may just be that the animal didn't move for a quarter of a second. To address this you can specify a minimum 'freeze' duration, such that the animal must freeze for at least this long before ANY-maze will count it as actually being in a freezing episode.

The value you enter can include units of ms (milliseconds) or s (seconds) and can include decimal places, so for example 1.5s and 1500ms are both valid. If you don't enter any units then seconds will be assumed.

The default value used is 2000ms, but you can change this to 0ms (so that even the shortest detectable 'freeze' will be scored) all the way up to 10s (in which case the animal has to freeze for 10 complete seconds before ANY-maze will consider it to be 'freezing').

Note that ANY-maze will score a freezing episode as starting at the time the animal started to freeze, not at the time at which it had been freezing for this minimum duration. For example, if the animal froze at time 10s and moved again at 12s, and the minimum freeze duration was 500ms, then at time 10.5s ANY-maze will decide that this really is a freezing episode. Since it knows the episode started at time 10s, it will report it as lasting for 2 seconds.

Note that you can alter the minimum freeze duration after a test has been performed. ANY-maze will simply reanalyse the data it collected during the test and update the freezing results accordingly.

Adjusting the freezing sensitivity

If you wish to, you can adjust the freezing detection system's sensitivity so that it is more or less strict in terms of how much movement it allows while still considering the animal to be freezing. Setting the sensitivity higher means that even a small movement will be detected and cause ANY-maze to consider the animal to NOT be freezing - so high sensitivity means the animal has to be very freezing (indeed at the highest sensitivity setting even breathing can be too much movement), while low sensitivity means it can move a little more. We recommend that you use the standard sensitivity setting and only adjust it if you need to.

Note that you can adjust freezing sensitivity after a test has been performed. ANY-maze will simply reanalyse the data it collected during the test and update the freezing results accordingly.

The difference between freezing detection and immobility detection

Immobility detection uses the ANY-maze tracking system to determine how much of the animal is moving. This works very well for determining whether the animal is staying in the same place, or even for determining if it is not moving its body very much, but it lacks the precision required to determine whether the animal is freezing.

© Copyright 2003-2026 Stoelting Co. All rights reserved

ANY-maze help topic T0359