and

Humans outperform, within limits, most algorithmic solutions in spatial optimization tasks, such as in the Euclidean Travelling Salesman Problem (TSP). It has been recently suggested that people follow simple, perceptual heuristics – like preferring closed shapes or following the boundaries – rather than applying computationally expensive algorithms. In the present paper, human navigational heuristics were simulated in standard 10-nodes TSPs with different levels of difficulty. Within an agent-based model, an artificial travelling agent evaluates a disutility function at each step with respect to the distance to potential object, memory of previous routes taken, and the distance from home. The number of steps in each tour was recorded and solutions were analyzed. We conclude that human navigational heuristics can lead to near-optimal solutions in tasks like the TSP, and that solutions modelled herein are both cognitively feasible and efficiently robust to a variety of circumstance