Based on animal long-term and short-term memory radial foraging techniques (or LMRFT and SMRFT), we devise a modelling approach that could capture the foraging behaviours of animals. In this modelling, LMRFT-based optimal foraging paths and SMRFT-based ones are constructed with respect to different levels of foraging strategies. Then, by a devised structural metric, we calculate the structural distance between these modelled optimal paths and the hypothetical real foraging paths taken by agents. We sample 20 foods positions via a chosen bivariate normal distribution for three agents. Then, we calculate their Euclidean distance matrix and their ranked matrix. Using LMRFT-based or SMRFT-based optimal foraging strategies, the optimal foraging paths are created. Then, foraging strategies are identified using optimal parameter learning techniques. Our results, based on the simulated foraging data, show that LMRFT-based foraging strategies for agent 1,2 and 3 are 3, 2 and 5, i.e., agent 3 is the most intelligent one among the three in terms of radial level. However, from the SMRFT-based perspective of strategies, their optimal foraging strategies are 5,5 and 2, respectively, i.e., agent 1 is as intelligent as agent 2 and both of them have better SMRFT-based foraging strategies than agent 3.
Keywords: LMRFT-based and SMRFT-based cognition; density-driven forage; experimental design; foraging radial level; structural metric.