Positional Inference in Rhesus Macaques Greg Jensen 1,2,5 , Vincent P. Ferrera 1,2,4 , and Herbert S. Terrace 3,4 1 Dept. of Neuroscience, Columbia University 2 Zuckerman Mind Brain Behavior Institute, Columbia University 3 Dept. of Psychology, Columbia University 4 Dept. of Psychiatry, Columbia University 5 Corresponding author (email: greg.guichard.jensen@gmail.com) ABSTRACT Understanding how organisms make transitive inferences is critical to understanding their general ability to learn serial relationships. In this context, transitive inference (TI) can be understood as a specific heuristic that applies broadly to many different serial learning tasks, which have been the focus of hundreds of studies involving dozens of species. In the present study, monkeys learned the order of 7-item lists of photographic stimuli by trial and error, and were then tested on “derived” lists. These derived lists combined stimuli from multiple training lists in ambiguous ways. We found that subjects displayed strong preferences when presented with novel test pairs. These preferences were helpful when test pairs had an ordering congruent with their ranks during training, but yielded consistently below-chance performance when pairs had an incongruent order relative to training. This behavior can be explained by the joint contributions of transitive inference and another heuristic that we refer to as “positional inference.” Positional inferences play a complementary role to transitive inferences in facilitating choices between novel pairs of stimuli. The theoretical framework that best explains both transitive and positional inferences is a spatial model that represents both the position and uncertainty of each stimulus. A computational implementation of this framework yields accurate predictions about both correct responses and errors for derived lists. Keywords: derived list, transitive inference, positional inference, serial learning, symbolic distance effect Serial learning refers to the ability of an organism to learn the underlying order of a set of items and to exploit that knowledge to make inferences about their relative position in that order (Terrace, 2010, 2012). As it is usually defined, such learning arises from experience and feedback, rather than from a comparison of the sensory features of the stimuli. For example, to order a collection of stones from smallest to largest, a subject could compare the size of the stones directly without reasoning about the stones as an ordered set. By contrast, learning the order of the Latin alphabet cannot be explained by mere stimulus comparisons. Nothing about the names of letters or the symbols that represent them suggests that Z has an obvious “lastness.” Its position is thus arbitrary relative to sensory properties, as is its close cousin Zeta being the 6 th item of the Greek alphabet. Serial learning is necessary for such arbitrary orderings to be encoded. Many simple organisms are able to organize their behavior without relying on serial learning at all. For example, single-celled organisms may display complex and adaptive behaviors in response to changing chemical gradients. However, the overwhelming majority of complex organisms that have been tested are able to behave in accord with implied orderings that are not signaled by explicit spatial or temporal cues (Jensen, 2017). Such behavior often requires the ability to make inferences about transitive serial relationships. Since the first non-human demonstration of transitive inference in squirrel monkeys (McGonigle and Chalmers, 1977), evidence of such inferences has been reported in dozens of species.