In this study, we developed a novel spontaneous recognition task to show that rats are able to associate objects to particular locations in an environment where these
locations can only be uniquely identified by combining landmark and odour cues. While similar experiments have assessed the ability of rats to perform configural processing (Eacott and Norman, 2004; Norman and Eacott, 2005), this is the first to require the use of contextual information to disambiguate directional information within a task.
The ‘location-odour’ task
While based upon the ‘object-place-context’ or ‘what-where-which’ task of Eacott and Norman (2004), the task described here differs in that the identity of the objects themselves are consistently identical and thus not important. With this in mind, this
‘location-odour’ task can be better aligned to the ‘context-place’ or ‘which-where’ task of Easton et al. (2010) where subjects were only required to associate a place with the context it is located in.
The ‘location-odour’ task takes fundamental principles from the spontaneous
recognition task, exploiting rats’ tendency to explore novel aspects of their environment (Ennaceur and Delacour, 1988). The findings from the final paradigm (Experiment 1D) showed that when one object was displaced at test, creating a visually symmetrical environment, the rats showed a clear preference (demonstrated by a significant increase in exploration time and re-exploration score) for objects in a novel location-odour configuration over an object in a familiar location-location-odour configuration. This provides strong evidence that the two olfactory contexts can provide orienting cues that are just as salient as visual information, and that behaving animals can integrate this context to encode the location of objects and resolve directional ambiguity.
With the preliminary experiments in mind, where animals familiar with the context box demonstrated a clear discrimination of the displaced and non-displaced objects in a single day paradigm, the three pilot experiments aimed to ascertain the level of familiarity needed for naïve animals to demonstrate the same level of discrimination.
Between the first and second pilot, animals were only exposed to additional experience with the context box environment; but between the second and third pilot experiment, the same animals were exposed to even further experience with the environment and a texture-enriched set of objects. Thus, the discrimination results from the third pilot could have been driven by either novelty (as the objects could be treated as new) or extended training. As it would not have been possible to disentangle these two
87 changes without additional pilot experiments in naïve animals, both changes were carried over into the final paradigm.
This type of task with free exploration demonstrates an animal’s natural behaviour rather than any trained behaviour or behaviour resulting from aversion or demand. As the compartments were distinguishable by the locations of the objects relative to landmarks during habituation, it is assumed that there is no demand on the animal to pay attention to the local odour cues for navigation but as per latent learning (Tolman, 1948) these should be processed regardless. Once the available cues were degraded to odour alone, the rat was able to use the specific odour contexts to recall the location-odour configuration successfully – this demonstrates that rats do indeed passively encode olfactory information without demand, and that olfactory information can be used as an orienting source especially under directional ambiguity. This fits well with neural data that found the hippocampus can substitute for missing visual information by using olfactory spatial context information to facilitate synaptic plasticity and enable spatial information encoding (André and Manahan-Vaughan, 2013).
In addition to indicating increasing familiarity, the learning trend over the initial days of the full paradigm may reflect increasing stability and depth of the animals’ spatial representation. Barry et al. (2007) found that the scale of grid cells recorded in the mEC expands coincidentally with the animal being introduced to a novel environment, but that this expansion then returns to baseline with three or four days of repeated experience. This suggests that spatial representations of environments become more stable with increasing experience, and could support the reasoning that naïve animals in the pilot experiments did not strongly react to the spatial displacement because their representation of the location-odour configuration was incomplete.
Although learning was noted within days, between days there was a noticeable effect that habituation from the previous day partially reversed after the 24-hour rest period.
From observation of the animals’ behaviour, this effect may not have been due to a reversal of learning but more to do with dishabituation or novelty detection after elapsed time. Given that there is no goal-directed behaviour or reward in this task, all learning being done is latent (Tolman, 1948) and thus it is more likely that the animals just initially explored more due to a curiosity of being in different surroundings (despite them being familiar).
Other interpretations
The contexts driven purely by odour in this experiment are different to conventional descriptions of context. This discussion so far has assumed that the distinct odours in
88 each compartment of the apparatus would create two distinct contexts, but there is a possibility that the box could be treated as a single context with the odours acting as orienting cues similar to the cue card landmarks. Recording of grid cells in a two-compartment connected environment (as in Skaggs and McNaughton (1998) (Figure 3.3A) showed that there was an experience dependent change in the grid cell
representation of these spaces: initially, the grid firing patterns were dominated by local environmental cues and displayed replicated patterns in the two compartments; but with experience, grid firing patterns formed a single continuous representation that spanned the environment as a whole (Carpenter et al 2015). This transition suggests that, within a 2 week period (15 sessions at a rate of 1 per day), grid cells adjust their firing to produce a globally coherent representation of the space as a whole rather than its composite compartments. In experiment 1D of this thesis, the animals were trained for only 5 sessions at a rate of one per day; thus with the timescale set by Carpenter et al. (2015) it would take an additional 10 sessions for grid cells to reach a single
continuous representation of this apparatus so at the point of test, the compartments would still have been treated as separate.
It must also be noted that as no control experiment was carried out in the two-compartment context apparatus without the use of any odours, it remains unproven that odour acts as the disambiguating cue. However, it is highly unlikely that the animals could resolve the spatial displacement at test without odours given the visual symmetry of the local environment from the point of view of the animal and the
absence of any global cues or experimenter-influenced processes (i.e. initially placing the rat consistently in one particular compartment) (Rosenthal and Fode, 2007).
In summary, the present study shows that rats can use olfactory context cues to create complex conjunctions of stimuli (in this case, location-odour) and resolve directional ambiguity of a visually symmetrical environment. A novel aspect of this experiment was that animals had two methods of discriminating context: by odour or by directional orientation, where the orientation was informed by the odour. In other words, the influence of odour on the compartment discrimination could have been direct, or indirect via the head direction system. The next two experiments explore this issue.
Experiment 2 records place cells to determine whether they remapped or repeated their fields between compartments, and if they repeated them, whether the fields reversed orientation between compartments. Experiment 3 manipulated the head direction system directly, by inactivating the anterior thalamic nuclei, to see whether rats could make the discrimination under a disruption of the head direction signal.
89