Download to read offline
Spatial Learning by mice on a 3D task 3
- 1. Spatial Learning bymice on a Three-dimensional radial maze Reference Memory task Benjamin James ABSTRACT Navigating strategies and the processes of exploration of rodents through two- dimensional space has been widely studied. However the real world is three- dimensional and little is known about how three-dimensional spaces are encoded and navigated in animals and humans. Using a Three-dimensional radiolarian maze task, which consists of a central spherical core from which 14 arms are projected in all directions, we hope to test Reference memory (Long term memory) in Three- dimensional space. Mice are required to explore the maze and retrieve food from the ends of the baited arms without missing out or revisiting baited arms they have already been too. During the reference memory task only a few of the arms will be baited. From previous findings we hope to show that during a three-dimensional task mice do not confuse the arms in the spatial learning task. This will give us an insight into the encoding of three-dimensional space and whether mice can navigate the vertical and horizontal parts of the task in their representation of space.
- 2. 2 Introduction Navigating through theworld is vital to the survivalof animals. Neural encoding of the world has survivaland evolutionary advantages. This ability requires the perception and encoding of spatial cues associated with important locations within the environment, cues such as food locations. It is thought that the hippocampus is central to the cognitive map and spatial mapping of complex space. Research has shown that animals, especially rodents, are well adapted to processing information abouttheir currentposition and the relative position of important objects and food sources. Currentwork supports the theory that animals hold an internal mechanismand representation of space. This is often referred to as the cognitive map (Jeffery et al. 2012). Previous Studies have focused on spatial navigation in two-dimensional environments. However the real world is Three-dimensional, with animals moving in a horizontaland vertical plane. Moving through three-dimensional space has additional computational requirements, which requires complex perception and encoding of spatial cues. Animals have to move againstgravity, which increases energy demands in three-dimensional movements compared to two-dimensional movements. The additional requirement for cognitively mapping height and horizontaldistance in three-dimensional environments, as opposed to planar environments, is much greater. The ability to synthesiseknowledgeof spatial cues and locations in three- dimensional space, rather than two-dimensionalspacewould provide more in depth information to the animal. This would aid in the precision of navigating around local environments and finding food sources. Ithas been suggested fromcurrent neural studies in rodents that vertical and horizontalspaces are encoded differently (Jeffery et al. (2011). The separation of vertical and horizontalinformation has also been discovered in fish (Holbrook & Burt de Perera, 2009).
- 3. 3 This would suggestthatvertebrate use a dual coded cognitive map for encoding vertical and horizontalplanes and that representation of space is flat in the plane of locomotion (Jeffery et al. 2012). Itis thought that the hippocampus in humans has a major role in spatial navigation and memory for life events. It is hoped fromstudies into the spatial navigational systems and mapping of three-dimensional spaces by rodents, we will be able to understand episodic memory (The memory for life events) in humans, in which it is thought a spatial framework system is used. Understanding episodic memory will help us understand the things that go wrong with it in many neurodegenerative conditions such as Alzheimer’s disease(Jeffery et al. 2014). Our study aims to see how complex Three-dimensional spaces are represented and used in navigation by mice. In our experiment we will be conducting a simple 3D navigation task, in which mice will explore and receive food rewards thatwill be placed at vertical and horizontal coordinates. We will be using a radiolarian maze to test the mice with. This is an unfamiliar environment, which the mice will haveto navigate and cognitively map to find food rewards. We will be focusing on reference memory (Long term memory) in mice. Although our study will primarily be focusing on reference memory, mice will have to use working memory (short-termmemory), to remember wherethey havebeen during the task and which arms the food reward is on and not on. We expect to see over the courseof five weeks mice making fewer mistakes, learning wherethe food rewards arelocated and only visiting these arms, with the number of errors decreasing and the mice eventually making no errors, or the number of errors plateauing for a number of days during the courseof the experiment. Prior to starting the research placement I had to read about the subject. This was so I understood whatwe would be trying to understand during the experiment and the currentknowledge surrounding spatialencoding. From Dr. Kate Jeffery’s lab page (http://www.ucl.ac.uk/jefferylab/research) I read
- 4. 4 about the researchcarried out at the lab and the aims of this research. I also read a review of currentresearch into spatial navigation in a 3D world that gave me an overview of research surrounding this area. My supervisor gavemesome usefularticles that gaveme a basic understanding aboutthe subject. Oneof the papers that I read was about Anisotropic encoding of three-dimensionalspace by place cells and grid cells (Jeffery et al. (2011)). This articlegave me a basic understanding of research into spatial encoding in rodents and enabled me to understand the aims of spatial encoding research. These articles were found online so I could read them. This helped me in writing my report and meant that I could understand the true aims of the experiment and whatwe were hoping to achieve. I also looked into some of the meanings of the scientific languageused in the literature and asked my supervisor aboutany words thatI was unsureabout. Methodology We would be using a 14 arm three-dimensional Radiolarian mazeto test reference memory. In this study only 12 of the arms would be recorded. We would bait 6 arms with condensed milk. Condensed milk was used as the food reward and not food pellets. The food needed to be a treat to ensuremice would want to eat the reward. 8 male mice would be used on the maze. We used 8 mice so we could compareperformances and ensure wehad enough mice to collect enough data to work outaverages and draw conclusions. We trained the mice during habituation for 5 days beforewe started the real experiment. This was to ensurethe mice would be comfortableto carry out the task and got used to the experimental environment. We had to weigh and handle the mice on the first day of the habituation phaseof the experiment. Handling continued on the second day. On the Third day of habituation mice were placed onto a restricted diet to maintain mice within 85-90% of freefeeding body weight.
- 5. 5 We then handledmice in the roomwe would be carrying outthe experiment in as partof habituation. We did this on day two of habituation. On the remaining three days we placed the mice on the radiolarian maze we would be using during the experiment. There was no food reward (Condensed milk) on the maze at this point. This was to get the mice used to the environment, so the mice would be happy to explore the radiolarian maze when it came to the real experiment. Mice explored the maze for 10 minutes each. We took records of the mice using a recording sheet. After each trial we input data into an excel spreadsheet. The mice were on a 12-hour light/dark cyclewith stimulated dawn at 23:30 and simulated dusk at 11:30. Allmice were trained during the dark cycle between 12:30 and 15:00. Each trial took around an hour (12:30-1:30) with the mice being given an hour break beforethe second trial started (2:30- 3:30). Between trials raw data was input into an Excel spreadsheet, to work out averages and analyseresults for each trial and day. This was important to do to ensureall data was input correctly and so we could see the progress of results between trials and days. As we were handling mice we needed to wear protective clothing to prevent contamination frompotential diseases and preventthe development of an allergy. We woregloves, facemask, overshoes, head cap and gown. We also had to wash are hands after each session as a safety precaution. During the experiment six of the 12 arms werebaited during each trial. The position of the baited arms was randomly assigned to each mouse; this was to ensure that the upper and lower arms on the radiolarian maze were equally baited (Three arms fromthe upper halve, three arms fromthe lower halve). Trials would last 5 minutes each or until all the baited arms had been visited and the rewards received. We conducted 2 trials a day. Onetrial started at 12:30 (during dark cycle) and once we had completed the firsttrial, mice weregiven a break and we then started the second trial at 2:30. Two trials were carried out instead of one so wecould have as many repeats as possibleand to compare performances between the firstand second trial. The maze was rotated by
- 6. 6 180 degrees everytrial to control for any cues there may be within the maze. Mice werescored on working memory errors (revisits to arms), reference memory errors (a visit to a non-baited arm) and visits to a baited arm. The order of arm visits and time taken to complete the maze were also noted. Two experimenters scored the recordings to make sure all recordings were constantto improvereliability. A video camera was also used that took a clip of each training session in case any visits were missed. Once wehad collected our data we placed our scores into an excel spreadsheetfor raw reference memory scores and day trials. The scores we collected enabled us to work outthe averages for each trial for total visits, omission (the number of baited arms missed out), commission (total visits - (6 - Omission), referencememory errors, revisits, time and Reference memory errors as a percentage of the total number of visits and working memory errors as a percentage of total visits. We also listed the order in which the arms werevisited. From this we created line graphs to presentresults. Mice would be trained on this task for five weeks or until the number of reference memory errors as a percentage of total visits remained constant for at least 3 days. Apparatus We used a 14-armthree-dimensionalradial armmaze (See figure 1 and 2). Only 12 of the arms would be used in the study. The top arm(0) and the bottom arm (-1) wereexcluded fromthe reference memory task. The spherewas 30cmin diameter with 14 evenly spaced cylindrical arms protruding fromthe sphere. The arms were3.5cmin diameter. A three- dimensional maze was used instead of a two-dimensionalmaze. A three- dimensional maze would be a more realistic representation of the real world (the real world is three-dimensional).
- 7. 7 Bandages were wrappedaround the body and arms of the maze; this would enable the mice to be able to climb and the maze with ease and not fall off. The maze was placed into an empty rack and suspended with nylon wire. Arms would be baited with condensed milk. Condensed milk would act as the food reward. This was placed onto the head of pins (Only six pins were baited at any one time). We would be using 8 male mice in our study that would be placed onto a restricted diet two days into the habituation phase of five days. This is to maintain the mice within 85-90% of theweight the mice were on when on unrestricted diet. We used Mice because they are lightweight natural climbers that would easily manage to explore the maze. Other rodents such as rats would be too large to be placed onto the maze. Figure 1-Radiolarian maze Figure 2-Radiolarian maze with labels 1-12 0 3 2 5 4 61 7 11 12 8 109 -1
- 8. 8 Results Results wereentered intoan Excel spreadsheet as below. We then worked out the averages for each Trial and the standard error, so wecould create line graphs to presentour results that included error bars. Averages and standard errors werealso worked out for each day. This was also done for the probetrials. Using IBMSPSS Statistics 21 we conducted repeated measures ANOVA Statistical analysis for both trial, day and probetrial. This showed that there had been statistically significant learning across total visits, omission (the number of baited arms missed out), commission (total visits - (6 - Omission), referencememory errors, revisits, timeand Reference memory errors as a percentage of the total number of visits and working memory errors as a percentage of total visits, by all mice on the radiolarian maze over the courseof the 50 trials (25 days). Any resultthat was below p=≤ 0.05 was seen as statistically significant. For the reference memory task we weremostly interested in the reference memory errors as a percentage of the total number of visits. We conducted 50 trials before the percentage of reference memory errors remained around 31-±3% (Below chance levels of 56%) for 3 consecutivedays, showing no improvement. Over the courseof the 25 days of trials the reference memory errors as a percentage of the total number of visits
- 9. 9 decreased from 54.46%to 30.00%(F (24,168) =5.82, p<. 001). Reference memory errors decreased from5.75 to 3.38 over the 25 days of trials (See Fig. 12). The time taken for the mice to complete the task decreased from296.94 seconds to 148.25 seconds (F (24,168) =10.51, p=<.001) over thecourseof the 25 days of trials (See Fig. 14). Working memory errors as a percentage of total visits decreased from19.43% to 6.18% (F (24,168)= 2.78, p<.001) over the courseof the 25 days of trials (seeFig. 16). Omission decreased from2.19 to 0.00 (F(24,168) =11.07, p<.001) over the 25 days of trials (See Fig.17). Totalvisits remained around 9.88 ±4.00 (F(24,168) =2.38, p=.001)over the25 days of trials (See Fig 11). Revisits decreased from 2.13 to 0.81 over the 25 days of trials (See Fig. 15). Commission decreased from6.06 to 4.06 over the 25 days of trials (See Fig. 18). Two probetrials wereconducted where no arms werebaited to see if the mice would visit the previously baited arms to prove the mice had learnt over the 25 days of trials. Repeated measures ANOVA was used to analyse the firstday of trials, the last day of trials and the two probe trials to confirmif there had been significant referencememory learning. Fromthe probetrials the analysis confirmed that over the 25 days there had been significant learning for the reference memory errors as a percentage of total visits. Between the first day and last day there was a significant decrease in reference memory errors as a percentage of total visits of 23.69%. Between the first day and probeday there was a decrease of 27.67% (seeFig. 19). This was significantaccording to repeated measures ANOVA, F(2,14) =45.27, p<.001. The results from repeated measures ANOVA would indicate that there had been significant learning by all the mice over 25 days of trials (Two trials a day). This would supportthe idea that mice can hold and store a visual representation of food sources in complex three-dimensionalspace over a long period of time.
- 10. 10 Figure 3. Average numberof total visits (Trial). Figure 4. Average working memory errors as a percentage of the total number of visits (Trial) Fig 5. Average of Reference memory errors as a percentage of the total visits (Trial). Fig 6. Average For the time taken to complete the task (Trial). Trial Graphs
- 11. 11 Fig.7 The average numberof revisits (Trial). Fig 8. Average Omission (Trial). Fig 9. Average Commission (Trial). Fig 10. Average of Reference memory Errors (Trial).
- 12. 12 Day Graphs Fig. 11. AverageTotal Visits (Day). Fig 12. Average for Reference memeory errors as a % of total visits. (Day) Fig 13. Average for Reference memeory errors (Day). Fig 14. Average for Time (Day).
- 13. 13 Fig 15. Average Revisits(Day). Fig 16. Average % Working Memory Errors (Day). Fig 17. Average Omission (Day). Fig 18. Average Commission (Day).
- 14. 14 ProbeTrial Graph Evaluation The findingsof this study highlight the ability of mice to locate and representpositions of food rewards within a three-dimensionalspace. However, further research would need to be conducted into the mechanisms of three-dimensional representations to further understand the cognitive map within the brain. This could be found out by electrophysiology studies of rodents to measurethe electrical properties of cells and tissues thought to be involved in spatial navigation such as place or grid cells. Variations of the radiolarian maze could be used to further investigate this. The radiolarian maze could have been adapted to include more or less arms or wecould have used a radiolarian maze of a different size, or a mazewith a different three-dimensional shape, such as a cube (seeFig. 20). When placing the food reward (Condensed milk) on the arms, the condensed milk would often drop off the pinheads. To improve the design of the radiolarian maze wecould have placed food rewards onto lollipop sticks or used a food reward that is less likely to drip off the end of the pins. Fig 19. Average Reference Memory Errors (Probe day).
- 15. 15 Other adaptations ofthe maze could be to design a small platform onto which food could be placed, thus preventing food rewards fromfailing off. Ear buds wereused to apply the condensed milk to the pinheads, as the condensed milk easily stayed on the ear buds and could easily be applied. We could have conducted our experiment differently by comparing Two- dimensional and three-dimensionalmazes to see differences in the way mice learn spatial tasks and the speed and ease at which they learn them. The study could have been conducted with other Rodents such as rats or we could haveused other mammals such as cats or dogs to make comparisons aboutspatial navigation. Further studies would haveto be conducted using a range of vertebrate, to confirmthe findings about spatial navigation in rodents and the cells that are involved. A larger radiolarian maze could havebeen used with more obstacles as this would be a more realistic representation of complex space and give a greater insight into spatial navigation in three-dimensional space. Overallthe experiment was conducted well. Over the courseof the 5 weeks of training, the mice did learn the location of all the baited arms and all successfully completed the radiolarian maze. This gave us an insight into the ability of mice to navigate the position of objects in three-dimensional space. Variations of the radiolarian maze might give us greater insight into the ability of mice to navigate three-dimensional environments. Theuse of electrophysiology experiments would have given us an understanding of the cells that are being used in navigational memory and cognitive mapping. I have learnt about the cells that are believed to be responsiblefor navigational memory such as grid cells and place cells. I havealso been given insight in the ability of mice to be able to learn the locations of food rewards.
- 16.
- 17. 17 Conclusions Using the Three-dimensionalradiolarian arm maze, we could see whether mice could represent locations in complex three-dimensional spaces, locating food rewards and holding this spatial memory over time. The mice learnt the reference memory task, learning food reward locations and holding these representations over long periods of time. The mechanisms behind this ability are still unclear. Itis hoped fromresearch into spatial navigation and its mechanisms that we can understand whathappens when this goes wrong in patients suffering from neurodegenerativediseases such as Alzheimer’s. Three-dimensional spatial navigationalstudies haveshed light on areas of the brain that are affected by neurodegenerative decline and the ways in which spatial navigation and memory are affected. There are declines in navigational skills in normal ageing with patients with dementia. This decline is a result of structuraland functional alterations in the neuralnetwork (Lithfous et al. 2013). Spatial navigational studies have given insights into the way the brain maps its surroundings and haveshown thatnavigational training programs can improvespatial performances in navigational tasks with patients suffering fromdementia. Further animal studies need to be conducted before we can fully understand the physiologicalmechanisms that are responsiblefor spatial navigation and mapping in the brain. Three-dimensional and two- dimensional studies in rodents as well as electrophysiology experiments have helped us understand moreabout spatial mapping and navigation.
- 18. 18 Appendix Buzsaki, G., Moser,E.I (2013) 'Memory, navigation and theta rhythmin the hippocampal-enthorhinalsystem', nature neuroscience, Vol16, no.2, pp. 130-138. References Hayman, R., Verriotis, M.A., Jovalekic, A., Fenton, A.A, Jeffery, K.J (2011) ‘Anisotropic encoding of three-dimensional spaceby place cells and grid cells’, nature neuroscience, Vol. 14, no.9, pp.1182-1188. Holbrook, R.I. and Burtde Perera, T. (2009) Separateencoding of vertical and horizontalcomponents of spaceduring orientation in fish. Animal Behaviour, 78(2), 241-245. Lithfous et al. (2013) 'spatialnavigation in normalaging and the prodromal stage of Alzheimer's disease: Insights fromimaging and behavioralstudies', Ageing research reviews, Vol. 12,Issue1, pp. 201-213. Jovalekic A, Hayman R, Becares N, Reid H, Thomas G, Wilson J, Jeffery KJ (2011) Horizontalbiases in rats’ useof three-dimensional space. Behavioral Brain Research, 222: 279-288. Jeffery et al. (2012) Navigating in a 3D world (online). Available at: http://www.ucl.ac.uk/jefferylab/publications/2013_Jeffery_BBS_preprint.p df (Accessed 4th August2014). Jeffery et al. (2014) Research (Online). Availableat: http://www.ucl.ac.uk/jefferylab/research (Accessed 4th August2014)
- 19. 19 Bibliography Grobéty, MC. &Schenk, F. (1992) ‘Spatiallearning in three-dimensional maze’, Animal behavior, Vol. 43, no.6, pp.1011-1020. Hayman, R., Verriotis, M.A., Jovalekic, A., Fenton, A.A, Jeffery, K.J (2011) ‘Anisotropic encoding of three-dimensional spaceby place cells and grid cells’, nature neuroscience, Vol. 14, no.9, pp.1182-1188. Holbrook, R.I. and Burtde Perera, T. (2009) Separateencoding of vertical and horizontalcomponents of spaceduring orientation in fish. Animal Behaviour, 78(2), 241-245. Lithfous et al. (2013) 'spatialnavigation in normalaging and the prodromal stage of Alzheimer's disease: Insights fromimaging and behavioralstudies', Ageing research reviews, Vol. 12,Issue1, pp. 201-213. Jovalekic A, Hayman R, Becares N, Reid H, Thomas G, Wilson J, Jeffery KJ (2011) Horizontalbiases in rats’ useof three-dimensional space. Behavioral Brain Research, 222: 279-288. Muller, R. (1996) ‘A Quarter of a Century of Place Cells’. Neuron, Vol.17, 979-990. Jeffery et al. (2012) navigating in a 3D world (online). Available at: http://www.ucl.ac.uk/jefferylab/publications/2013_Jeffery_BBS_preprint.p df (Accessed 4th August2014). Jeffery et al. (2014) Research (Online). Availableat: http://www.ucl.ac.uk/jefferylab/research (Accessed 4th August2014)
- 20. 20 Acknowledgements With thanks tothe Nuffield foundation and Emma Newall for providing the placement. Many thanks to Kate Jeffery Lab UCL and Jonathan Wilson for agreeing to allow me to work with them in their lab for my placement and giving me an insight into behavioral neuroscience. The project has confirmed to me that I wantto pursuea career in neuroscienceand has shown me how a real lab operates and the exciting research that goes on at the UCL institute of Behavioural neuroscience. This has opened my mind to considering a career in neuroscienceresearch.