Marco Vasconcelos

Transitive inference in non-human animals: An empirical and theoretical analysis

Transitive inference has long been considered one of the hallmarks of human deductive reasoning. Recent reports of transitive-like behaviors in non-human animals have prompted a flourishing empirical and theoretical search for the mechanism(s) that may mediate this ability in non-humans. In this paper, I begin by describing the transitive inference tasks customarily used with non-human animals and then review the empirical findings. Transitive inference has been demonstrated in a wide variety of species, and the signature effects that usually accompany transitive inference in humans (the serial position effect and the symbolic distance effect) have also been found in non-humans. I then critically analyze the most prominent models of this ability in non-human animals. Some models are cognitive, proposing for instance that animals use the rules of formal logic or form mental representations of the premises to solve the task, others are based on associative mechanisms such as value transfer and reinforcement and non-reinforcement. Overall, I argue that the reinforcement-based models are in a much better empirical and theoretical position. Hence, transitive inference in non-human animals should be considered a property of reinforcement history rather than of inferential processes. I finalize by shedding some light on some promising lines of research.

Pro-sociality without empathy

Empathy, the capacity to recognize and share feelings experienced by another individual, is an important trait in humans, but is not the same as pro-sociality, the tendency to behave so as to benefit another individual. Given the importance of understanding empathys evolutionary emergence, it is unsurprising that many studies attempt to find evidence for it in other species. To address the question of what should constitute evidence for empathy, we offer a critical comparison of two recent studies of rescuing behaviour that report similar phenomena but are interpreted very differently by their authors. In one of the studies, rescue behaviour in rats was interpreted as providing evidence for empathy, whereas in the other, rescue behaviour in ants was interpreted without reference to sharing of emotions. Evidence for empathy requires showing that actor individuals possess a representation of the receivers emotional state and are driven by the psychological goal of improving its wellbeing. Proving psychological goal-directedness by current standards involves goal-devaluation and causal sensitivity protocols, which, in our view, have not been implemented in available publications. Empathy has profound significance not only for cognitive and behavioural sciences but also for philosophy and ethics and, in our view, remains unproven outside humans.

Darwin's "tug-of-war" vs. starlings '" horse-racing": how adaptations for sequential encounters drive simultaneous choice

Charles Darwin aided his private decision making by an explicit deliberation, famously deciding whether or not to marry by creating a list of points in a table with two columns: “Marry” and “Not Marry”. One hundred seventy-two years after Darwin’s wedding, we reconsider whether this process of choice, under which individuals assign values to their options and compare their relative merits at the time of choosing (the tug-of-war model), applies to our experimental animal, the European Starling, Sturnus vulgaris. We contrast this with the sequential choice model that postulates that decision-makers make no comparison between options at the time of choice. According to the latter, behaviour in simultaneous choices reflects adaptations to contexts with sequential encounters, in which the choice is whether to take an opportunity or let it pass. We postulate that, in sequential encounters, the decision-maker assigns (by learning) a subjective value to each option, reflecting its payoff relative to background opportunities. This value is expressed as latency and/or probability to accept each opportunity as opposed to keep searching. In simultaneous encounters, choice occurs through each option being processed independently, by a race between the mechanisms that generate option-specific latencies. We describe these alternative models and review data supporting the predictions of the sequential choice model.

How costs affect preferences: experiments on state dependence, hedonic state and within-trial contrast in starlings

We examined how starlings, Sturnus vulgaris, adjust preferences to retrospective sunk costs in either time or work. Ideal decision-makers disregard sunk costs, but under some circumstances animals, like humans, prefer normally costlier rewards when they do not have to pay the costs. We argue that a possible explanation is state-dependent valuation learning (SDVL). The argument is that subjective value (hedonic state or fitness) is a decelerated function of energetic state, and energetic costs displace the subject’s state towards conditions where the function is steeper and the subjective value of rewards is enhanced. We compared SDVL with within-trial contrast (WTC). WTC says that experiencing a food reward brings subjects to a baseline, constant, hedonic state, so that being in a more negative prereward mood causes a bigger positive displacement. SDVL focuses on energetic costs, while WTC focuses on any emotional dimension. We carried out three experiments that differed in the salience and dimension of the cost (time or work) and found overvaluation of costlier outcomes when the retrospective cost involved work (locomotory activity) but not when it involved only delays. When work was used as a cost, the effects of cost on preference showed strong stochastic transitivity. Our findings are consistent with studies in which energetic state was altered by deprivation rather than cost, and may help to explain why studies of WTC have had unreliable outcomes.

Irrational choice and the value of information.

Irrational decision making in humans and other species challenges the use of optimality in behavioural biology. Here we show that such observations are in fact powerful tools to understand the adaptive significance of behavioural mechanisms. We presented starlings choices between probabilistic alternatives, receiving or not information about forthcoming, delayed outcomes after their choices. Subjects could not use this information to alter the outcomes. Paradoxically, outcome information induced loss-causing preference for the lower probability option. The effect depended on time under uncertainty: information given just after each choice caused strong preference for lower probability, but information just before the outcome did not. A foraging analysis shows that these preferences would maximize gains if post-choice information were usable, as when predators abandon a chase when sure of the prey escaping. Our study illustrates how experimentally induced irrational behaviour supports rather than weakens the evolutionary optimality approach to animal behaviour.

Deprivation level and choice in pigeons: a test of within-trial contrast.

Within-trial contrast has been proposed as a mechanism underlying preferences for stimuli that follow relatively more aversive events over stimuli that follow less aversive events. In this study, we manipulated deprivation level to test within-trial contrast predictions. In Experiment 1, pigeons encountered two discriminative stimuli, one presented when they were deprived and the other when they were prefed. When later given a choice between the two stimuli, pigeons strongly preferred the stimulus encountered when deprived, independently of their deprivation level at test. In Experiment 2, pigeons learned two simultaneous discriminations, one when deprived and the other when prefed. Here, subsequent tests between the two S+ or the two S− stimuli revealed no consistent preferences. These contrasting findings suggest that differential aversiveness is necessary but not sufficient to induce preferences via within-trial contrast.

Starlings uphold principles of economic rationality for delay and probability of reward

Rationality principles are the bedrock of normative theories of decision-making in biology and microeconomics, but whereas in microeconomics, consistent choice underlies the notion of utility; in biology, the assumption of consistent selective pressures justifies modelling decision mechanisms as if they were designed to maximize fitness. In either case, violations of consistency contradict expectations and attract theoretical interest. Reported violations of rationality in non-humans include intransitivity (i.e. circular preferences) and lack of independence of irrelevant alternatives (changes in relative preference between options when embedded in different choice sets), but the extent to which these observations truly represent breaches of rationality is debatable. We tested both principles with starlings (Sturnus vulgaris), training subjects either with five options differing in food delay (exp. 1) or with six options differing in reward probability (exp. 2), before letting them choose repeatedly one option out of several binary and trinary sets of options. The starlings conformed to economic rationality on both tests, showing strong stochastic transitivity and no violation of the independence principle. These results endorse the rational choice and optimality approaches used in behavioural ecology, and highlight the need for functional and mechanistic enquiring when apparent violations of such principles are observed.

Choice in multi-alternative environments: A trial-by-trial implementation of the Sequential Choice Model

The sequential choice model (SCM) proposes that latencies to accept options presented alone can be used to predict preferences between these options when they are presented simultaneously. SCM has been proposed and tested in experiments where only two alternatives were present. To further challenge the model, we trained and tested European starlings (Sturnus vulgaris) in an environment with a background of four alternatives differing in delay to reinforcement. Unexpected binary choices between the six possible pairs of alternatives were used to assess preference. The models predictions of the strength of preference roughly corresponded to the birds choices for each of the six choice situations. More importantly, a trial-by-trial test of the model correctly predicted 84% of all individual choice trials.

Acquisition versus steady state in the time-left experiment

To test some predictions of scalar expectancy theory (SET) for the time-left procedure, we performed one experiment with two conditions. In Condition A, pigeons were exposed to two fixed-interval schedules, a fixed-interval (FI) 30s and an FI 60s, each associated with a distinct key and presented on a separate trial. Subsequently, during test trials, the FI 60-s key was illuminated and then after T = 15, 30 or 45 s the FI 30-s key also was illuminated. The main issue was how choice between the two keys varied with T. Condition B replicated Condition A with different FI parameters and T values. The results showed that (a) contrary to SETs predictions, preference changed reliably with testing, which suggests that learning took place during the test trials; (b) within each test trial, pigeons revealed an almost exclusive preference for one of the keys, and (c) at steady state pigeons behaved in the same way as rats. Because SET could not account for these findings we advanced a new descriptive model of performance for the time-left task. The model fit the data well.

Cognitive mechanisms of risky choice: is there an evaluation cost?

We contrast two classes of choice processes, those assuming time-consuming comparisons and those where stimuli for each option act independently, competing for expression by cross censorship. The Sequential Choice Model (SCM) belongs in the latter category, and has received empirical support in several procedures involving deterministic alternatives. Here we test this model in risky choices. In two treatments, each with five conditions, European starlings (Sturnus vulgaris) faced choices between options with unpredictable outcomes and risk-free alternatives. In the delay treatment the five conditions involved choices between a variable option offering two equiprobable delays to reward and a fixed option with delay differing between conditions. The amount treatment was structurally similar, but amount of reward rather than delay was manipulated. As assumed (and required) by the SCM, latency to respond in no-choice trials reflected each options richness with respect to the background alternatives, and, crucially, preferences in simultaneous choices were predictable from latencies to each option in forced trials. However, we did not detect reliable differences in response times between forced and choice trials, neither the lengthening expected from evaluation models nor the shortening expected from the SCM.