Behavioral aspects of spatial competition between red wood ants (Formica aquilonia) and six mass species of Carabidae were studied in field and laboratory experiments. We showed that red wood ants essentially influence spatial distribution of ground beetles on their common territories. Transplantation experiments suggest that in newly established ants' settlements stronger forms of interrelations arise than in old stable colony. To examine the ability of beetles to avoid collisions with ants we used two experimental techniques. In laboratory, we tested carabids ability to avoid a clash in a Y-shaped labyrinth containing an active tethered ant in one section. In field experiments we compared quantitative characteristics of movements (such as crookedness of individual trajectories, speed of movement, the time spent on stops) for beetles placed close to ants foraging routes and on ant-free plots. All beetles studied displayed a clear tendency to learn, that is, to modity their behavior in order to avoid collisions with ants. Species that exhibited best parameters of learning were closer to ants by their size and characteristic movement, namely, Pterostichus oblogopunctatus and P. magus. Beetles' stereotyped behavioral tactics can be considered universal for avoiding collisions with any subject (for instance, with an ant) of a certain size and speed of movements. A set of tactics in the labyrinth included: (1) attempts to round the ant; (2) turns away after touching the ant with antennae; (3) turns away without a contact; (4) avoidances of a dangerous section; (5) stops near the ant with the antennae hidden. Comparing pairwise difference between four species shows that beetles use species-specific preference for definite combinations of tactics. Effective learning allows carabids to penetrate into ant foraging territory and partly avoide interference competition. It seems that red wood ants are not inclined to learn to avoid collisions with competing carabid species. Instead, they recognize an "enemy's image" and selectively attack relatively small predatory carabids rather than herbivorous species. Experiments with dummy beetles suggest that ants react for several hierarchically organized key characteristics of competitors such as speed of movement, dark color, bilateral symmetry, protuberances (legs, antennae), and smell. Among "professional" groups in ant family, guards and hunters react to beetles aggressively, whereas aphid tenders ignore them. Sophisticated combination of flexible and innate behavioral patterns enables insects to share territories and interact in the mode of relatively mild, spatial, competition instead of predation. Eliciting sets of hierarchically organized features of competitors that govern adjustment of spatial distribution in insects' community enable us to integrate ideas of classic and cognitive ethology.