Refining training of non-human primates using automated home room training systems
Lead Research Organisation:
University College London
Department Name: Institute of Neurology
Abstract
Traditionally, the process of training non-human primates (NHPs) on cognitive tasks takes place in the laboratory. This involves removing the NHP from the comfort of their home room, and social group, for training each day. NHPs then perform tasks in a confined testing chair, placing restrictions on their mobility for the duration of the session and also limiting the types of behaviours that can be explored. This process can increase stress levels for the NHP. Furthermore, the laboratory-based training schedule is dictated by the experimenter, with NHPs typically only trained for 1-2 hours/day, 4-5 days/week, prolonging the training process and exacerbating these welfare concerns.
To overcome these issues, we recently developed the Mymou (Greek for "monkey", pronounced my-moo) system, a low-cost home room training system for NHPs (Butler & Kennerley, 2018). The wireless device runs continuously and automatically all-day including weekends, allowing NHPs to perform tasks in their own home room at their own leisure and comfort. This eliminates any need to place restrictions on the NHP's movement, including head restraint, or to remove them from the security of their home room, for the training process. Furthermore, the system uses a camera to snap a 'selfie' of the NHP for each trial they start. This selfie is run through a custom monkey facial recognition algorithm we developed that is capable of accurately identifying NHPs with >99% accuracy. This enables NHPs in the same cage to be trained without needing to be separated from one another. Mymou is therefore a significant refinement for the training process.
Furthermore, the constant availability of Mymou in the home room enables NHPs to complete many more trials per week relative to training in the lab, refining the training process by reducing the overall time the NHP spends under protocol. This enables researchers to train NHPs on more diverse and sophisticated tasks, thereby allowing the experimenter to obtain higher quality data whilst potentially reducing the total number of NHPs needed to address their scientific aims.
Having now successfully refined NHP training procedures at University College London, this project will repeat this in the other UK NHP neuroscience research centres (Oxford, Cambridge, and Newcastle Universities). For the 8 end users we will develop a personalised cognitive task(s) relevant to their research, spanning a range of tasks exploring attention, memory, learning and decision-making processes. We will then install these personalised systems in each centre, and provide training on how to use the device. This will enable home room training, and the refinements this provides, for 26-36 NHPs across the 3 centres.
An important aim of this project is to also encourage adoption of home room training by the wider NHP community. This project will produce a suite of standardised training and cognitive tasks that are used across the varied domain of NHP behavioural research, allowing users the flexibility to explore how their NHPs perform on different tasks. This will help optimize individual training regimens and also aid in assigning NHPs to projects they are best suited for.
All of these resources will be made publicly available online in our actively maintained repository, providing a rich foundation of information to help NHP researchers get started with home room training. Furthermore, encouraging home room training across the UK will generate a critical mass of users to help encourage the wider community. The results from this project will be publicised through many channels (e.g. conferences, press releases, scientific papers), which will help encourage the adoption of home room training worldwide, therefore providing a refinement and reduction to NHP research across the entire community.
To overcome these issues, we recently developed the Mymou (Greek for "monkey", pronounced my-moo) system, a low-cost home room training system for NHPs (Butler & Kennerley, 2018). The wireless device runs continuously and automatically all-day including weekends, allowing NHPs to perform tasks in their own home room at their own leisure and comfort. This eliminates any need to place restrictions on the NHP's movement, including head restraint, or to remove them from the security of their home room, for the training process. Furthermore, the system uses a camera to snap a 'selfie' of the NHP for each trial they start. This selfie is run through a custom monkey facial recognition algorithm we developed that is capable of accurately identifying NHPs with >99% accuracy. This enables NHPs in the same cage to be trained without needing to be separated from one another. Mymou is therefore a significant refinement for the training process.
Furthermore, the constant availability of Mymou in the home room enables NHPs to complete many more trials per week relative to training in the lab, refining the training process by reducing the overall time the NHP spends under protocol. This enables researchers to train NHPs on more diverse and sophisticated tasks, thereby allowing the experimenter to obtain higher quality data whilst potentially reducing the total number of NHPs needed to address their scientific aims.
Having now successfully refined NHP training procedures at University College London, this project will repeat this in the other UK NHP neuroscience research centres (Oxford, Cambridge, and Newcastle Universities). For the 8 end users we will develop a personalised cognitive task(s) relevant to their research, spanning a range of tasks exploring attention, memory, learning and decision-making processes. We will then install these personalised systems in each centre, and provide training on how to use the device. This will enable home room training, and the refinements this provides, for 26-36 NHPs across the 3 centres.
An important aim of this project is to also encourage adoption of home room training by the wider NHP community. This project will produce a suite of standardised training and cognitive tasks that are used across the varied domain of NHP behavioural research, allowing users the flexibility to explore how their NHPs perform on different tasks. This will help optimize individual training regimens and also aid in assigning NHPs to projects they are best suited for.
All of these resources will be made publicly available online in our actively maintained repository, providing a rich foundation of information to help NHP researchers get started with home room training. Furthermore, encouraging home room training across the UK will generate a critical mass of users to help encourage the wider community. The results from this project will be publicised through many channels (e.g. conferences, press releases, scientific papers), which will help encourage the adoption of home room training worldwide, therefore providing a refinement and reduction to NHP research across the entire community.
Technical Summary
Traditionally, training non-human primates (NHPs) on cognitive tasks occurs in the laboratory, which involves removing the NHP from their home room and social group. NHPs then perform tasks in a confined testing chair, restricting their mobility for the session. This process can increase stress levels for the NHP, and leads to inefficient task training, often requiring over a year to teach an NHP a complex task.
To overcome these limitations, we developed an automated, low cost, home-room training system for NHPs. The fully wireless device runs continuously all-day including weekends, allowing NHPs to perform tasks at their leisure. The system uses facial recognition to accurately identify the NHP, eliminating the need to separate NHPs from their social group and home room for training. Mymou significantly increases the amount of training NHPs complete per week, which allows for more ambitious and impactful experiments to be undertaken in each NHP, and expedites training.
This project will install home room systems in multiple labs across the UK (Oxford, Cambridge, and Newcastle Universities). We will develop a battery of standardized and personalised tasks for each end user, and provide on-site training to use the system, allowing for the immediate refinement in the training of 26-36 NHPs. The tasks provided will explore motor, attention, memory, decision-making and social behaviours, in both marmosets and rhesus macaques. We will collate data across tasks and labs to examine factors that facilitate learning to further optimise NHP training.
The project is fully open source, and all the different tasks and enclosure designs will be made publicly available in our actively maintained online repository. The results will be presented at conferences, in scientific papers, and through invited visits to our facility. This will encourage the international NHP community to adopt home-room training, therefore helping to refine and reduce NHP research worldwide.
To overcome these limitations, we developed an automated, low cost, home-room training system for NHPs. The fully wireless device runs continuously all-day including weekends, allowing NHPs to perform tasks at their leisure. The system uses facial recognition to accurately identify the NHP, eliminating the need to separate NHPs from their social group and home room for training. Mymou significantly increases the amount of training NHPs complete per week, which allows for more ambitious and impactful experiments to be undertaken in each NHP, and expedites training.
This project will install home room systems in multiple labs across the UK (Oxford, Cambridge, and Newcastle Universities). We will develop a battery of standardized and personalised tasks for each end user, and provide on-site training to use the system, allowing for the immediate refinement in the training of 26-36 NHPs. The tasks provided will explore motor, attention, memory, decision-making and social behaviours, in both marmosets and rhesus macaques. We will collate data across tasks and labs to examine factors that facilitate learning to further optimise NHP training.
The project is fully open source, and all the different tasks and enclosure designs will be made publicly available in our actively maintained online repository. The results will be presented at conferences, in scientific papers, and through invited visits to our facility. This will encourage the international NHP community to adopt home-room training, therefore helping to refine and reduce NHP research worldwide.
Planned Impact
This project will provide considerable refinement and reduction to NHP neuroscience research, including both a direct and immediate impact to the UK centres and an indirect but wide reaching impact to the wider community.
Direct impact
This project will equip 8 end-users across all of the UK NHP neuroscience centres (Oxford, Cambridge, and Newcastle Universities) with bespoke home room training devices. The systems will be used directly for home room training of 26-36 NHPs instead of laboratory-based training, and it is likely that many more NHPs in the future will also be trained in this manner.
NHPs find separation from the security of their home room, or from their social group, stressful, and this project will eliminate this necessity for the training of 26-36 NHPs on their respective cognitive tasks. Home room training places no restrictions on the NHPs mobility and they are free to come and go from the device as they please. This therefore ensures that food/fluid rewards are available to the NHP when they desire them most.
Furthermore, we found home room training to be 560% more efficient compared to lab training (as it runs all day every day fully unsupervised), greatly refining the length of the training process. This increased efficiency also enables research to progress at a quicker pace, or NHPs to learn more complicated experiments addressing multiple scientific aims in the same experiment. This means the scientific impact of each NHP will be greater, which can lead to a reduction in the overall number of NHPs required for neuroscience research.
Finally, the Mymou system allows NHPs to begin learning behavioural tasks shortly after arriving at the facility. This means training can proceed without the need for any implants, head restraint or even food/fluid control. Indeed, we completed training on an extremely complicated associative learning task all within the home room prior to performing any surgery on the NHP. As implants can potentially be a source for infection or cause other adverse effects, it is a significant refinement to overall health if training can be completed before implants are required.
Indirect impact
This project will see the development of a suite of cognitive tasks that are commonly used across the diverse field of NHP research. 8 custom enclosures designs will also be generated to attach Mymou onto a variety of different home room designs. All of this will be made available on an online repository, enabling users (either pre-existing or new) to easily access and start using Mymou. The repository will also serve as a community where people can ask for help or guidance from more experienced users with any issues they may have. This will therefore encourage adoption of the Mymou system, and the benefits this brings to NHP research, across the international NHP research community.
In addition, we will promote the benefits and success of home room training achieved with this project at scientific conferences attended by the international NHP community and wider cognitive neuroscience field. We will also promote the project through press releases to the wider audience, and host visits from any researchers keen to find out exactly how home room training works. This should therefore help encourage for the uptake of the Mymou system and home room training, and refinements and reductions to the welfare of NHPs, across the international NHP community.
This project will therefore result in a considerable refinement and reduction to NHP research. Both directly to end users that span the major neuroscience research users in the UK, and indirectly to the wider NHP community by facilitating and encouraging update of home room training.
Direct impact
This project will equip 8 end-users across all of the UK NHP neuroscience centres (Oxford, Cambridge, and Newcastle Universities) with bespoke home room training devices. The systems will be used directly for home room training of 26-36 NHPs instead of laboratory-based training, and it is likely that many more NHPs in the future will also be trained in this manner.
NHPs find separation from the security of their home room, or from their social group, stressful, and this project will eliminate this necessity for the training of 26-36 NHPs on their respective cognitive tasks. Home room training places no restrictions on the NHPs mobility and they are free to come and go from the device as they please. This therefore ensures that food/fluid rewards are available to the NHP when they desire them most.
Furthermore, we found home room training to be 560% more efficient compared to lab training (as it runs all day every day fully unsupervised), greatly refining the length of the training process. This increased efficiency also enables research to progress at a quicker pace, or NHPs to learn more complicated experiments addressing multiple scientific aims in the same experiment. This means the scientific impact of each NHP will be greater, which can lead to a reduction in the overall number of NHPs required for neuroscience research.
Finally, the Mymou system allows NHPs to begin learning behavioural tasks shortly after arriving at the facility. This means training can proceed without the need for any implants, head restraint or even food/fluid control. Indeed, we completed training on an extremely complicated associative learning task all within the home room prior to performing any surgery on the NHP. As implants can potentially be a source for infection or cause other adverse effects, it is a significant refinement to overall health if training can be completed before implants are required.
Indirect impact
This project will see the development of a suite of cognitive tasks that are commonly used across the diverse field of NHP research. 8 custom enclosures designs will also be generated to attach Mymou onto a variety of different home room designs. All of this will be made available on an online repository, enabling users (either pre-existing or new) to easily access and start using Mymou. The repository will also serve as a community where people can ask for help or guidance from more experienced users with any issues they may have. This will therefore encourage adoption of the Mymou system, and the benefits this brings to NHP research, across the international NHP research community.
In addition, we will promote the benefits and success of home room training achieved with this project at scientific conferences attended by the international NHP community and wider cognitive neuroscience field. We will also promote the project through press releases to the wider audience, and host visits from any researchers keen to find out exactly how home room training works. This should therefore help encourage for the uptake of the Mymou system and home room training, and refinements and reductions to the welfare of NHPs, across the international NHP community.
This project will therefore result in a considerable refinement and reduction to NHP research. Both directly to end users that span the major neuroscience research users in the UK, and indirectly to the wider NHP community by facilitating and encouraging update of home room training.