To navigate from place to place, an agent needs to have knowledge of the spatial structure of the world and a strategy for using that knowledge to choose a route. In this talk, I will describe work from my lab that investigates how these core elements of navigation are implemented by the human mind/brain. First, I will describe studies that explore the organization of spatial knowledge ("cognitive maps"). Participants in these experiments were first familiarized with virtual environments containing multiple objects and then scanned with fMRI while recalling the spatial relationships between the objects. Analysis of behavior and fMRI activation patterns revealed that participants' cognitive maps were not simple Euclidean reference frames—rather, they had a rich structure that reflected the environments' segmented and hierarchical organization. Second, I will describe a study that explores the algorithms that people use to navigate to a goal. Participants in this study were scanned with fMRI while performing a "taxi-cab" task that required them to flexibly navigate to multiple possible locations within a virtual city. We found that people's behavioral choices could be explained by a predictive coding (successor representation) model, and that fMRI signals in the hippocampus and neocortical spatial memory regions tracked the different computational components of the model. Together these findings illuminate the representations that people form of environmental spaces and the algorithms they use to navigate to goal locations within these spaces.

I will present data on spatial navigation in preclinical and clinical Alzheimer’s disease (AD) patients. Specifically, they will illustrate how spatial navigation/orientation changes are highly specific to underlying AD pathophysiology and can manifest as one of the earliest symptoms of the disease. These changes have been demonstrated not only in clinical populations but also, significantly, in genetic (APOE) and sporadic (biomarker) at-risk-of-AD individuals before the onset of memory problems. I will further highlight how these spatial navigation/orientation changes can significantly impact patients’ daily movements and their outdoor behaviour. The aim is to convey that spatial disorientation is an under-recognized symptom in AD, which allows to study spatial navigation changes within a lesion model while at the same time having clear clinical relevance for monitoring preclinical disease and ensuring patient safety later in the disease progression.

Much of the analysis of human geography relates to the relationship between people and the space around them. Geography shapes individual perceptions and decisions, and in turn informs in the emergence of social and spatial phenomena that impact our everyday lives. Yet the analytical processes we use in understanding human behaviour in geographic space are relatively limited, compared to the wider literature around humans and spatial cognition. This talk will outline some of the research that explores the way in which geography shapes human behaviour. This research builds on unprecedented datasets that link characteristics of the built environment to human behaviour at population scales. It also explains how geographic data and predictive models are being adapted to better account for spatial and behavioural features, to improve predictions of social systems where human spatial behaviour is important.

To survive in a changing world, animals often need to suppress an obsolete behavior and acquire a new one. This process is known as reversal learning. The neural mechanisms underlying reversal learning in spatial navigation have received limited attention and it remains unclear what neural mechanisms maintain behavioral flexibility. Here, we extend a closed-loop simulator of spatial navigation and learning, based on spiking neural networks. In this model, activity of place cells and boundary cells are fed as inputs to action selection neurons, which drive the movement of the agent. Upon reaching the goal, behavior is reinforced with spike-timing-dependent plasticity (STDP) coupled with an eligibility trace which marks synaptic connections for future reward-based updates. We model a task with an ABA design, where the goal is switched between two locations A and B after 10 trials. The challenge is understanding how a biologically plausible spiking neural network maintains flexibility and its associated costs. To measure the agent's performance, we employed three methods: trial duration, proximity, and similarity between the agent's traversed trajectory and an ideal trajectory (DTW). All measurements were consistent. Agents using symmetric STDP excel initially on finding target A, but fail to find target B after the goal switch, persevering on target A. Either using asymmetric STDP, using many small place fields, or injecting short noise pulses to action selection neurons are effective in driving spatial exploration, which ultimately leads to finding target B. However, this flexibility comes at the price of slower learning and lower performance. Our work shows three examples of neural mechanisms that achieve flexibility at the behavioral level, each with different characteristic costs.