Reducing animal numbers in tasks of memory

Spontaneous recognition tasks are widely used as a measure of an animal's memory. This new apparatus allows reduced variability in this task, which results in a significant reduction in animals required. The nature of the apparatus lends itself to automation and to use in a variety of different tasks assessing different forms of memory reliant on different neural systems. Partners are sought to help develop and validate the new apparatus that allows reduced variability in spontaneous recognition tasks.

Through the CRACK IT Solutions scheme Dr Alex Easton partnered with GSK. With the Solutions funding awarded, the team adapted and verified the apparatus for use with mice, demonstrating a 50% reduction in the number of mice required per study (from 16- 20 mice to just eight). 

This project and its impacts are featured as a case study in the 2019 CRACK IT Review.

Spontaneous recognition tasks are used extensively in studies of Alzheimer’s Disease and other conditions (including normal ageing) where memory loss and decline in cognitive function are consequences of the condition. In these spontaneous recognition tests, the animals simply explore the environments (and objects) around them. Their preference for novelty means that any difference in exploration of different objects can be used as a measure for the animals’ memory of previously seen objects. These tests of object memory (or more complex memory of objects and places or objects and contexts) primarily use one trial per animal or when multiple trials are used they are carried out at a frequency of one trial per day. This results in slow data collection as well as large levels of variability in test animal behaviour which requires large numbers of animals to produce sufficient statistical power.

Current apparatus that allows multiple trials requires animals to shuttle backwards and forwards, with different test trials occurring in different locations in the apparatus. This means that only object recognition memory can be tested, not more complex tasks including memory of objects in particular locations. These additional tasks are important to understand the neural mechanisms of different types of memory, which are often differentially affected by ageing and disease processes.

Through an NC3Rs-funded PhD studentship, the team at Durham University led by Dr Alex Easton, have developed an apparatus that allows multiple trials by a single rat to be run in a single session with no handling between trials (Ameen-Ali et al., 2012). This enables tests on different trials to be carried out in the same physical location, and therefore spatial components of memory can be tested in addition to object memory. The device allows the local environment in which the test trials occurs to be changed between trials allowing the testing of even more types of memory, including (potentially) models of episodic memory - critical for understanding memory loss in ageing and disease processes. The equipment allows multiple trials by a single animal to be run in a single session with no handling between trials. This produces a significant reduction in animal numbers required on this task. In addition, the lack of handling reduces stress in the animals. This allows these tasks to be more widely used in experimental settings, with both greater throughput of data and more subtle changes in behaviour able to be observed.

The apparatus is a modification of the spontaneous recognition memory task and so is applicable for use in both research and drug development as a simple measure of an animal's memory ability. These tasks have been used in basic research to explore the neural mechanisms of different types of memory, but also in screening assays to better understand the effect of drugs on memory. In addition, the repeated trial nature of the behaviour now allows these well-defined tasks to be adapted for single cell recording in a way single trial per day versions of the task cannot be. In all of these situations, this novel apparatus significantly reduces (by around 50%) the number of animals used in this task, as well as allowing variations of the task to be used with similar reduced numbers.


  • Ameen-Ali KE, Eacott MJ & Easton A (2012). A new behavioural apparatus to reduce animal numbers in multiple types of spontaneous object recognition paradigms in rats. Journal of Neuroscience Methods 211(1): 66-76. doi:10.1016/j.jneumeth.2012.08.006.

In order for the apparatus to be more widely used further work is required to develop the technology from its current format (manually operated and only tested in rats) to an easily used format for all users (i.e. automated and useful for all rodents). This work needs collaborative input to develop and test a new version of the apparatus suitable for studies with mice and technical support to develop automation methods for a modified version of this apparatus. The automation of the apparatus would require mechanical, electrical and programming skills. The researchers have excellent mechanical facilities available to them but require support in developing suitable designs that suit automation for a variety of possible test situations. In addition, the apparatus needs to be validated for use in drug development and collaborations are sought with partners able to provide compounds to test this.

Recent work has suggested that using this novel apparatus animal numbers in this widely used task can be approximately halved, whilst maintaining the same statistical power. This comes from improved reliability of data through minimised handling and repeated measures from a single animal within a single session. The reduced handling serves to reduce stress in the animal, meaning the measure of behaviour is more reliable within animals. As more data is collected per animal, this drives the overall reduction in numbers. However, as the reduced stress also produces more reliable data this means smaller effect sizes are likely to be more readily observed. At the moment spontaneous recognition tasks of memory are often plagued by lack of replication or absence of significant effects due to variability in control animal data. This new apparatus enables researchers to overcome both these obstacles. Furthermore, automation of this novel apparatus may result in a system that can be attached to the home cage allowing animals to enter and perform trials when they choose to whilst still maintaining aural and olfactory contact with their cage mates. A broad literature search suggests that in the past five years this methodology could have reduced the use of animals in published studies by approximately 15,000.

Overview | Impact | Publication


To encourage wider adoption of the apparatus, especially within industry, Dr Alex Easton is keen to develop the spontaneous object recognition multiple trials apparatus from its current format (manually operated and only tested in rats) to an easily used format for all users (i.e. automated and useful for all rodents). Through the CRACK IT Solutions scheme, Dr Alex Easton partnered with GlaxoSmithKline (GSK, Shanghai) to further develop the apparatus and validate the technology using compounds provided by GSK.


The team has successfully tested mice in an adapted version of the original rat apparatus and demonstrated that it is possible to reduce by approximately 50% the number of animals required. Traditional single trial per day spontaneous recognition memory approaches requires 16 - 20 mice being tested for many days. However, the new apparatus generates highly reliable data on common tasks of memory (spontaneous object recognition and object location) using just eight mice running multiple trials consecutively without handling.

They have also demonstrated using a mouse model of Alzheimer's disease (TASTPM) that the equivocal results generated in standard one trial a day procedures using these mice could be due to anxiety in early trials which diminishes over the course of multiple trials. In our apparatus the TASTPM mice showed no significant impairment in common tasks of memory, contradicting some reports in the literature. This is an important result and suggests that the type of prolonged testing without handling provided by the apparatus the team has developed will provide more reliable data which is less prone to misinterpretation.

The combination of verifying the apparatus works for mice, and demonstrating that the results are more reliable and less susceptible to artefacts (such as anxiety) has completely changed the practice of working in their own laboratory. All behavioural studies they carry out using spontaneous tasks in both rats and mice now use a version of a continual trials apparatus. In addition, collaborations they have with other institutions are now utilising this approach.

The NC3Rs CRACK IT Solutions funding has not only enabled validation this approach, but it has also been instrumental in securing additional funding for Dr Easton to work with Campden Instruments to develop a semi-automated version of the apparatus for a commercial market. Having piloted this apparatus in Durham, prototype versions are now being readied for trial in other laboratories in the UK, focusing on habituating animals to the apparatus, running multiple apparatus simultaneously and better approaches to scoring behaviour. This further development is opening up the way to a wider scale adoption of the approach.


Chan M, Eacott MJ, Sanderson DJ, et al. (2018). Continual trials spontaneous recognition tasks in mice: reducing animal numbers and improving our understanding of the mechanisms underlying memory. Frontiers in Behavioral Neuroscience, 12: 214. doi:10.3389/fnbeh.2018.00214.

Chan M, Austen JM, Eacott MJ, et al. (2019) The NMDA receptor antagonist MK-801 fails to impair long-term recognition memory in mice when the state-dependency of memory is controlled. Neurobiology of Learning and Memory. 161: 57-62. doi: 10.1016/j.nlm.2019.03.006.