Edited by: Nadine Ravel, UMR5292 Centre de Recherche en Neurosciences de Lyon (CRNL), France
Reviewed by: Francoise Schenk, University of Lausanne, Switzerland; Rosamund Fay Langston, University of Dundee, United Kingdom
*Correspondence: Atomu Sawatari
†These authors have contributed equally to this work.
This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
Environmental enrichment (EE) via increased opportunities for voluntary exercise, sensory stimulation and social interaction, can enhance the function of and behaviours regulated by cognitive circuits. Little is known, however, as to how this intervention affects performance on complex tasks that engage multiple, definable learning and memory systems. Accordingly, we utilised the Olfactory Temporal Order Discrimination (OTOD) task which requires animals to recall and report sequence information about a series of recently encountered olfactory stimuli. This approach allowed us to compare animals raised in either enriched or standard laboratory housing conditions on a number of measures, including the acquisition of a complex discrimination task, temporal sequence recall accuracy (i.e., the ability to accurately recall a sequences of events) and acuity (i.e., the ability to resolve past events that occurred in close temporal proximity), as well as cognitive flexibility tested in the style of a rule reversal and an Intra-Dimensional Shift (IDS). We found that enrichment accelerated the acquisition of the temporal order discrimination task, although neither accuracy nor acuity was affected at asymptotic performance levels. Further, while a subtle enhancement of overall performance was detected for both rule reversal and IDS versions of the task, accelerated performance recovery could only be attributed to the shift-like contingency change. These findings suggest that EE can affect specific elements of complex, multi-faceted cognitive processes.
The ability to quickly formulate a consistent course of action based on salient events, as well as rapidly adjust behaviours to abrupt changes in contingencies is vital for survival. Often, relevant events happen in quick succession and thus require the ability to temporally resolve experiences. Further, the chronological order of these past events in and of themselves can provide vital information for choosing a correct series of actions.
The acquisition of stable rules (Wallis et al.,
Environmental enrichment (EE), which exposes subjects to enhanced levels of sensory stimuli, social interaction, and/or voluntary exercise (Hebb,
Behaviourally, EE has been shown to influence spatial memory acquisition and recall (Garthe et al.,
As a first step, an established memory assay, the Olfactory Temporal Order Discrimination (OTOD; Fortin et al.,
We found that EE improved acquisition of the OTOD task for all three temporal separations tested. Change point analyses on cumulative performance profiles of individual animals revealed that EE was associated with a significantly earlier upward transition in performance, particularly for middle and long intervals, indicating an expedited learning of task rules in these mice. Enriched groups also exhibited small but detectable performance improvements in two different versions of the same task with specifically altered task contingencies.
Together, these findings suggest that EE from birth can affect the use of multiple memory systems to better adapt to a rapidly and continuously changing environment.
This study was carried out in accordance with the recommendations of the Australian code for the care and use of animals for scientific purposes (editions 7/8), National Health and Medical Research Council (NHMRC) Guidelines, Animal Welfare Committee. The protocol was approved by Animal Ethics Committee (AEC) of the University of Sydney. All animals were housed in a single room at the University of Sydney Animal Housing Facility on a fixed 12/12 h light/dark cycle. All mice were housed in open-top cages and had
On arrival into our mouse colony, late pregnant C57BL6 mouse mothers (ARC, Western Australia) were randomly allocated into either enriched environment (EE), or standard environment (SE) housing conditions (see below). All experiments were performed on the offspring of these dams (Figure
Experimental timeline and task structure.
SE cages (12.5 × 30 × 11.5 cm) contained a red plastic igloo, paper chip pelleted bedding, and tissue for nesting. A single pregnant dam was placed in a given SE box.
EE cages were larger (12.5 × 45 × 30 cm) and, in addition to the components found in SE cages, featured a running wheel and a range of sensory stimuli including (but not limited to) cardboard tubing, half tissue boxes, rubber balls, scented cotton balls (e.g., artificial strawberry, vanilla and lemon scents), Velcro strips (adhered to cage walls to provide tactile stimuli), high contrast visual gratings, bell balls etc. These items were rearranged within the cage every second day. Further, late pregnant dams were pair housed in EE boxes to augment social interaction. Access to food and water was identical between mice placed in standard and enriched environments.
Litters remained in the same housing as their dams until weaning. At postnatal day (P) 21, the juveniles were placed into new cages maintaining their EE (23 mice from 2 litters: 5–8 animals per cage) and SE conditions (26 mice from 2 litters: 3–5 animals per cage). Behavioural testing commenced at P30 (Figure
On the evening prior to each behavioural session, food pellets were removed to encourage reward seeking behaviour during training/testing. Behavioural sessions took place every second day. Sunflower seeds were used as reward. Animals had
To evaluate memory for temporal order of a sequence of events, we exploited the natural proclivities that mice possess for digging and their aptitude for detecting and discriminating olfactory stimuli. Animals were first habituated to a testing arena (arena size: 60 × 60 × 40 cm), and subsequently trained to dig in sand-filled plastic cups for a buried sunflower seed reward linked to olfactory cues. A single odourant (0.01% cinnamon by weight mixed in sand) was used for this stage. This stage was repeated over two training days.
Mice were then exposed to four “weighted” sequence presentations in which five cups with different odours (0.01% odourant by weight; dry, powdered odourants included cardamom, clove, coriander, cumin, garlic, ginger, mustard, nutmeg, paprika, parsley, tea and turmeric), placed at one of two “foraging” locations within the arena (two adjacent corners of the enclosure). Odour cups presented earliest in the sequence contained four seeds, decreasing incrementally by a seed each cup, with the final two cups having no seeds in order to emphasise the importance of presentation order. Placement and order of odours presented were varied for each sequence presentation. Mice had 90 s to consume each reward. The subject was then removed from the arena for 30 s before being placed back in the enclosure with a new odour cup. All mice were given four complete sequences of five odours.
Subsequently, mice were presented with a sequence of five weighted (see above) odour cups followed by a choice test in which two of the odour cups (probe cups) from different time points in sequence were placed in both foraging corners. The subject had to choose the odour cup previously presented in the sequence to gain a reward (four seeds). Mice had 90 s to make a choice. The combination of a given five odour sequence with the choice test constituted a single trial. Each animal was exposed to five trials, each with randomised odour order and reward location; once on a span of three odours between choices (first and last odours: long distance), twice on a span of two odours (e.g., first and fourth or second and fifth; middle distance) and twice on a span of one odour (e.g., first and third; short distance). At the end of each trial, the subject was removed from the arena and placed in their home cage for 60 s before beginning the next trial.
This shaping procedure (including foraging, weighted sequence presentation alone, and weighted sequence presentation with choice test) was completed over four training days for each animal (Figure
Upon completion of shaping, mice were introduced to the actual OTOD task (Figures
After experiencing the five odours, mice were returned to the arena and presented with a choice test: two of the olfactory stimuli in probe cups, only one of which concealed a reward (four seeds). Animals that chose to dig in the cup containing the odour encountered earlier in the sequence were rewarded (correct choice). Foraging in the cup containing the more recently presented olfactory stimuli was deemed an incorrect response (incorrect choice; not rewarded) and the trial was terminated. The two probe cups were placed in the same two foraging corners within the arena used during shaping, with the rewarded cup assigned in a pseudo-random fashion (balanced for both sides) to avoid any spatial biases in choice selection. Placement of a paw within either cup constituted a choice made. On any given choice presentation, mice were allowed 90 s to find and consume the reward. If the subject did not make a choice in the time allotted (non-response), the probe was terminated and the animal removed from the arena (Figure
As during the shaping period, a trial consisted of one sequence of five odours and the subsequent probe. Any given discrimination problem was characterised by one of three different temporal “distances”: long (three odours apart; L); middle (two odours apart; M); short (one odour apart; S). On a training day (session), each animal received five consecutive trials (1 L, 2 M and 2 S) with an inter-trial interval of 60 s where the order of odor presentation in the sequence was changed (Figure
In order to investigate the effect of EE on cognitive/behavioural flexibility in the OTOD task, either reinforcement or stimulus contingencies were changed in the form of a Rule Reversal (RR) or an Intra-Dimensional Shift (IDS) respectively. To be rewarded, RR required animals to adapt their decision-making strategy to an inversion of the previously learned rule (i.e., choosing the odour experienced later in the sequence instead of that which was experienced earlier). The IDS required the animals to apply the learned rule to novel stimuli of the same olfactory modality (i.e., five new exemplar odours were used though the rules of the task did not change; Figure
A subset of each housing group (EE:
Due to the binary nature of the outcome, i.e., either a “correct” or “incorrect” choice, associations between housing condition, temporal span and performance were evaluated using general estimating equations (GEE; SPSS, IBM Corporation, NY, USA). A binomial distribution with a logit (logarithm of the odds ratio (OR): the probability of choosing to correct choice over the probability of choosing the incorrect choice) link function was used to model task acquisition, RR and IDS versions of OTOD. Associations between housing and performance were also modelled separately for each distance. Outcomes are presented as regression coefficients (b), standard errors in regression estimates (s.e.r.) and OR. OR values greater than one indicate significantly (
To further explore the dynamics of OTOD task acquisition, RR and IDS, a change point analysis was performed on the cumulative record of correct responses for each animal during each task type (Gallistel et al.,
The number of trials to the first change point indicating an upward shift in performance (which provides a measure of how quickly the animals exhibited a significant improvement in performance) and the pre- (Phase 1) and post- (Phase 2) change point slopes of the cumulative record (the correct response rate for a given epoch and therefore providing an indication of initial as well as asymptotic (post change point to end of trial period) performance levels) were calculated for each individual.
This was executed on the record of all trials across the entire OTOD task acquisition as well as for each temporal distance separately. For RR and IDS profiles, the change point analysis was only executed on the entire record since there were too few trials to reliably detect change points if separated by distance. An independent samples
Overall, both enriched and standard groups improved their performance over time for all three temporal separations tested, considered together or separately (Figure
Environmental enrichment (EE) affects overall acquisition of the OTOD task. Performance indicated by percentage correct over total choices for groups of EE (coloured) and standard environment (SE; grey scale) mice, all distances combined and plotted in accordance to OTOD task training schedule
Quantitative analysis revealed that EE mice exhibited higher performance scores compared to SE mice especially during earlier learning stages, suggesting that EE fosters improved task acquisition (Figures
When temporal separations were considered separately, EE mice also exhibited improved performance compared to SE mice for each separation (long:
The acquisition of novel discrimination tasks through trial-and-error learning can often be characterised by discrete, detectable and often abrupt transitions in performance levels across trials (Gallistel et al.,
When examining the cumulative sum of correct responses across the entire testing period, all animals exhibited at least one statistically significant change point, which marked the trial on which the performance of the animal improved most dramatically. EE mice required fewer trials to reach the first change point, indicating an accelerated upward transition in performance (Mann Whitney U-Test,
Initial OTOD acquisition is accelerated in EE mice.
When distances were considered separately only a portion of the animals exhibited a statistically significant change point (EE: 12/23 for long, 20/23 for middle, 18/23 for short; SE: 14/26 for long, 20/26 for middle, 21/26 for short). Nonetheless, when animals that demonstrated a change point were compared, upward transitions in performance occurred earlier for enriched animals compared to standard animals for long and middle, but not short distances (independent samples
In order to explore these differences further, the slopes of the cumulative performance plots before (Phase 1) and after (Phase 2) the identified change point were calculated. While the slopes were measurably different between the two phases, no housing effect was detected either for all trials together (Phase 1: EE = 0.55; SE = 0.51. Phase 2: EE = 0.85; SE = 0.83; mixed-model ANOVA: group (
Together these findings suggest that although both groups acquired and performed the task in a similar fashion, EE animals acquired the task more rapidly than their SE counterparts.
Survival is dependent on an animal’s ability to adapt to contingencies resulting from a rapidly and continuously changing environment. In an attempt to gain an overall impression of how enrichment can influence behaviour flexibility, the responses of mice from SE and EE cohorts trained on the OTOD were compared on one of two new conditions: (1) an inversion of the learned rule (Rule Reversal; RR) in which subjects were exposed to the original group of five odours presented in different sequences at the beginning of each trial as before, but were then required to identify the most recently presented scent to gain reward (SE:
Qualitative analysis revealed a dramatic difference in initial performance between the two new tasks. When examining performance for all subjects in each group, mice that underwent RR persevered with the previously learned obsolete rule during the initial stages of task re-acquisition (illustrated by “below chance” (~20%) mean performance levels exhibited by both housing groups across the first few trials; Figure
Enrichment affects immediate performance after both Intra-Dimensional Shifts (IDSs) and Rule Reversals (RR).
When performance was regressed on housing conditions, task type, and temporal distances, all three factors were shown to exert an influence on choice selection. Overall, EE mice were ~1.4 times more likely to choose the correct odour compared to SE cohorts (
When considered separately by task type, EE mice exhibited better recovery for both task variations. Enriched mice were more 1.36 times more likely to choose correctly in the RR version compared to SE cohorts (RR:
Together, these findings suggest that EE can instil advantages in choice selection even under conditions of uncertainty.
In order to better characterise the manner in which EE affected OTOD performance after a change in task contingencies, change point analyses were conducted on learning trajectories of individual animals for both RR and IDS cohorts.
All subjects exhibited at least one statistically significant upward transition (Figure
Enrichment accelerates performance recovery after an IDS but not a Rule Reversal.
No differences between housing groups were detected in performance slopes before and after the first upward change point for either the RR (Phase 1: EE = 0.29; SE = 0.23. Phase 2: EE = 0.69; SE = 0.63; repeated measures ANOVA: group (
These findings suggest that for the IDS task, performance improvements observed in EE mice were due to an expedited adjustment in behavioural strategy, manifesting as an application of the previously learned rule to a new set of exemplar stimuli.
The successful performance of the OTOD task is thought to require the processing of memory for temporal order or relative recency, which is mediated by the hippocampus (Fortin et al.,
Although the control housing condition is referred to here, and in the bulk of the relevant literature as “standard”, it should be noted that conventional laboratory housing conditions can be considered to provide a somewhat deprived sensory experience relative to what would be encountered by mice that are not held in captivity. Accordingly, the EE paradigm used in this study may be providing a more naturalistic environment for pups to develop (Arai et al.,
The mice in this study began their training while they were at a relatively immature stage, P30. Both the striatum and hippocampus, structures important for the acquisition and execution of the OTOD, begin to form perineuronal nets, extracellular structures associated with network consolidation and associated with the closing of heightened periods of neural plasticity during early development (critical periods), at roughly 3 weeks after birth (Brückner et al.,
As task acquisition continued for 2 months from around P30, training over this period in and of itself could have potentially provided enriching experiences for both EE and SE mice. Despite this, possible confound, we nevertheless observed a difference in acquisition during task engagement based on housing conditions. Indeed, it is possible that further changes due to EE from birth may have been masked due to the enriching effects of prolonged training in SE mice. Whether the observed, accelerated learning was due to a cumulative effect of enrichment from birth overlapping with task related training, or solely a result of early enrichment is not clear. Moreover, as training was initiated in adolescents, whether the effects we observed would be detectable in mice trained as adults is not known. Further experiments separating and isolating early enrichment from task acquisition in mature animals will be required to determine whether the observed differences in the current study can be uniquely attributable to the early enrichment period.
Given that mice have an innate preference for exploration of novelty (Smith et al.,
Although this study does not provide any direct evidence for the mechanisms underlying the improved performance observed in EE mice, it is possible that enrichment enhanced one of a number of sensory, learning and memory-associated, and decision-making processes to facilitate task acquisition. First, an enhanced ability to identify discrete odours may have facilitated OTOD learning. EE has been shown to improve performance on olfactory discrimination tasks (Mandairon et al.,
A number of mnemonic processes may also have been employed by the animals (such as relative recency, novelty/familiarity detection, episodic-like recall etc.) to learn and execute OTOD. Several lines of evidence have shown that spatial pattern separation is improved by enrichment and related processes (Creer et al.,
Successful execution of the task must have also engaged processes that link this relevant information to a course of actions that would lead to positive outcomes. The mapping of such associations involves the activation of a number of interacting brain regions, including the hippocampal formation (Fortin et al.,
Further, establishing appropriate rules of engagement for a particular task is potentially a dissociable process from being able to use this information to make appropriate decisions or execute a correct set of actions (Tsujimoto et al.,
The responses exhibited to the two modified versions of the OTOD task provides some insight into the degree to which some of these processes may have been differentially affected by EE. In the IDS version of the task, a single contingency change was implemented: a specific feature of the stimulus (i.e., odours) was changed while the rules governing successful completion of the task were maintained. EE mice exhibited expedited performance recovery under these conditions, even though initial, post-transition performance fell to chance levels. This suggests again that mice of both housing groups did not immediately utilise any innate preference for novelty, echoing the lack of novelty bias observed in the early stages of OTOD acquisition.
While the assignment of new olfactory exemplars in this task has been nominally identified as an “IDS”, as only one contingency change was tested in animals initially trained on a single set of odours, the experience of the subjects in this context would have been akin to the formation of a novel stimulus-reward association (as supported by the return of mean performance to chance levels). In light of this, the accelerated performance recovery as revealed by the change point analysis for EE mice may reflect an enhanced ability to update the cue-reward association to encompass the new task-related olfactory exemplars, possibly via enhanced medial prefrontal function (Tse et al.,
Conversely, in the rule reversal condition, while EE cohorts exhibited improved overall choice selection, consistent with previously reported changes in reversal learning for enriched mice (Zeleznikow-Johnston et al.,
The overall change in performance observed EE mice may instead reflect an improved ability to form new cue-action-outcome associations, consistent with augmented sensory/mnemonic processing implicated in the IDS results, a process that would have been initiated after abandoning the original rule. While the limited number of post-transition trials did not permit a closer examination of individual performance dynamics after the first upward transition, future studies should endeavour to explicitly examine how EE affects a subject’s ability to consolidate newly learnt associations upon discarding previously acquired knowledge; key processes that contribute to behavioural flexibility.
In conclusion, EE expedited the ability of mice to discriminate the temporal order of olfactory stimuli. This improvement appears to be related primarily to an earlier manifestation of learning on choice behaviour. EE animals also exhibited generally enhanced performance after a rule reversal, consistent with previous findings (Zeleznikow-Johnston et al.,
DR-H acquired the data. DR-H, TJB, CAL and AS contributed to the design of the work and interpretation of the data. TJB and AS performed the bulk of the analysis. DR-H, TJB and CAL contributed to the writing and editing of the manuscript. AS was responsible for writing, compiling and editing the final draft of the manuscript.
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
The authors would like to acknowledge the assistance and expertise of the Bosch Animal Behavioural Facilities staff in completing this work. No external funding was received to complete this work.