Do cleaner fish learn to feed against their preference in a reverse reward contingency task?

The ability to control impulsive behaviour has been studied in animals with a standard test in which subjects need to choose the smaller of two food items in order to receive the larger one (reverse reward contingency). As a variety of mammals that have been tested so far (mostly primates) have great difficulties to solve the task, it has been proposed that it is generally cognitively demanding. However, according to an ecological approach to cognition, a species’ ability to solve the task should not depend on its general cognitive abilities but on whether its ecology causes selective pressure on the ability to restrain foraging behaviour. We tested this hypothesis using the cleaner wrasse (Labroides dimidiatus), a fish species that feeds against its preference in nature when engaging in cleaning interactions with so called ‘client fish’. None of the eight tested individuals learned to choose a non-preferred item after 200 trials. In a subsequent test, one subject learned to respond correctly in a large or none contingency task (only the choice of the small food was rewarded). After a short re-experience treatment, this individual learned to solve the reverse reward task after 30 trials. In conclusion, we did not find support for the general idea that interactions with clients prepared cleaners to quickly solve a reverse reward test. However, the results suggest that the potential to solve a reverse reward contingency may not be restricted to mammals but could be present also in a fish species in which the problem of choosing a non-preferred food over a preferred one is an ever present challenge in nature.     

Is your choice my choice? The owners’ effect on pet dogs’ (Canis lupus familiaris) performance in a food choice task

This study investigates the influence of owners on their dogs’ performance in a food choice task using either different or equal quantities of food. Fifty-four pet dogs were tested in three different conditions. In Condition 1 we evaluated their ability to choose between a large and small amount of food (quantity discrimination task). In Condition 2 dogs were again presented with a choice between the large and small food quantity, but only after having witnessed their owner favouring the small quantity. In Condition 3 dogs were given a choice between two equally small quantities of food having witnessed their owner favouring either one or the other. A strong effect of the owner on the dogs’ performance was observed. In Condition 1 dogs as a group chose significantly more often the large food quantity, thus showing their ability to solve the quantity discrimination task. After observing their owner expressing a preference for the small food quantity they chose the large quantity of food significantly less than in the independent choice situation. The tendency to conform to the owner’s choice was higher when the dogs had to choose between equally small quantities of food (Condition 3) rather than between a large and a small one (Condition 2). These results provide evidence that dogs can be influenced by their owners even when their indications are clearly in contrast with direct perceptual information, thus leading dogs to ultimately make counterproductive choices.     

Transfer of self-control in black (Eulemur macaco) and brown (Eulemur fulvus) lemurs: choice of a less preferred food item under a reverse-reward contingency

When presented a choice between two food-type arrays of equivalent size under a reverse-reward contingency, black (Eulemur macaco) and brown (Eulemur fulvus) lemurs transposed their self-control abilities, acquired in a previous experiment, to significantly select the less-desired food item in order to gain access to the more desired one. However, when presented with the choice between two different food-type arrays in which the amount of the less desired food array was larger than the more desired one, large individual differences were revealed: Some subjects established a consistent rule favoring quality or quantity, whereas others exhibited various point of trade-off. These results show that lemurs seem to manage the task considering not only food quantity but also food quality.     

Squirrel monkeys (Saimiri sciureus) choose smaller food arrays: long-term retention, choice with nonpreferred food, and transposition

Squirrel monkeys (Saimiri sciureus) that had learned to reach toward 1 piece of food instead of 4 in a reverse-reward contingency were tested after an 8-month delay with no intervening relevant experiences. All monkeys except 1 continued to show inhibitory control by reliably reaching toward the smaller quantity, most of them doing so within 2 sessions. Performance was maintained when a low-preference food replaced prized foods as arrays and rewards. When the quantities 1 and 4 were replaced with different ones, there was strong evidence of transposition at group level, although individual differences in bias toward the smaller quantities became apparent. Individual differences in mastering the original task more than 8 months previously were quite stable, suggesting robustness in the operations required for this form of self-control.     

Colour versus quantity as cues in reverse-reward-competent squirrel monkeys (Saimiri sciureus)

To assess the relative salience of colour and quantity cues, squirrel monkeys previously trained to reach for the smaller of two quantities of food in a reverse-reward contingency task received colour discrimination training. After initial failure to discriminate between two colours of dots under a differential reinforcement regime, they learned the task when the S- colour was associated with zero reward. The monkeys then showed good retention on the original reverse-reward task of 1 versus 4 with pairs of dots presented in S+ or S- colours. However, on "mismatch" trials of 1S- versus 4S+ , only 2 of 4 monkeys tested showed a preference--1 monkey chose based on quantity, the other based on colour. Individual differences and the possible roles of overshadowing and blocking are discussed.     

How the great apes (Pan troglodytes, Pongo pygmaeus, Pan paniscus, and Gorilla gorilla) perform on the reversed contingency task: the effects of food quantity and food visibility

S. T. Boysen and G. G. Berntson (1995) found that chimpanzees performed poorly on a reversed contingency task in which they had to point to the smaller of 2 food quantities to acquire the larger quantity. The authors compared the performance of 4 great ape species (Pan troglodytes, Pongo pygmaeus, Pan paniscus, and Gorilla gorilla) on the reversed contingency task while manipulating food quantity (0-4 or 1-4) and food visibility (visible pairs or covered pairs). Results showed no systematic species differences but large individual differences. Some individuals of each species were able to solve the reversed contingency task. Both quantity and visibility of the food items had a significant effect on performance. Subjects performed better when the disparity between quantities was smaller and the quantities were not directly visible.     

Colour versus quantity as cues in reverse-reward-competent squirrel monkeys (Saimiri sciureus)

To assess the relative salience of colour and quantity cues, squirrel monkeys previously trained to reach for the smaller of two quantities of food in a reverse-reward contingency task received colour discrimination training. After initial failure to discriminate between two colours of dots under a differential reinforcement regime, they learned the task when the S? colour was associated with zero reward. The monkeys then showed good retention on the original reverse-reward task of 1 versus 4 with pairs of dots presented in S + or S? colours. However, on “mismatch” trials of 1S? versus 4S + , only 2 of 4 monkeys tested showed a preference—1 monkey chose based on quantity, the other based on colour. Individual differences and the possible roles of overshadowing and blocking are discussed.     

Comparisons of Online Reading Paradigms: Eye Tracking, Moving-Window, and Maze

This study compares four methodologies used to examine online sentence processing during reading. Specifically, self-paced, non-cumulative, moving-window reading (Just et al. in J Exp Psychol Gen 111:228–238, 1982), eye tracking (see e.g., Rayner in Q J Exp Psychol 62:1457–1506, 2009), and two versions of the maze task (Forster et al. in Behav Res Methods 41:163–171, 2009)—the lexicality maze and the grammaticality maze—were used to investigate the processing of sentences containing temporary structural ambiguities. Of particular interest were (i) whether each task was capable of revealing processing differences on these sentences and (ii) whether these effects were indicated precisely at the predicted word/region. Although there was considerable overlap in the general pattern of results from the four tasks, there were also clear differences among them in terms of the strength and timing of the observed effects. In particular, excepting sentences that tap into clause-closure commitments, both maze task versions provided robust, “localized” indications of incremental sentence processing difficulty relative to self-paced reading and eye tracking.     

The role of awareness in anticipation and recall performance in the Hebb repetition paradigm: implications for sequence learning

Sequence learning has notably been studied using the Hebb repetition paradigm (Hebb, 1961) and the serial reaction time (SRT) task (Nissen & Bullemer, Cognitive Psychology 19:1–32, 1987). These two paradigms produce robust learning effects but differ with regard to the role of awareness: Awareness does not affect learning a repeated sequence in the Hebb repetition paradigm, as is evidenced by recall performance, whereas in the SRT task, awareness helps to anticipate the location of the next stimulus. In this study, we examined the role of awareness in anticipation and recall performance, using the Hebb repetition paradigm. Eye movements were monitored during a spatial reconstruction task where participants had to memorize sequences of dot locations. One sequence was repeated every four trials. Results showed that recall performance for the repeated sequence improved across repetitions for all participants but that anticipation increased only for participants aware of the repetition.     

Increasing stimulus intensity does not affect sensorimotor synchronization

When people synchronize taps with isochronously presented stimuli, taps usually precede the pacing stimuli [negative mean asynchrony (NMA)]. One explanation of NMA [sensory accumulation model (SAM), Aschersleben in Brain Cogn 48:66–79, 2002] is that more time is needed to generate a central code for kinesthetic-tactile information than for auditory or visual stimuli. The SAM predicts that raising the intensity of the pacing stimuli shortens the time for their sensory accumulation, thereby increasing NMA. This prediction was tested by asking participants to synchronize finger force pulses with target isochronous stimuli with various intensities. In addition, participants performed a simple reaction-time task, for comparison. Higher intensity led to shorter reaction times. However, intensity manipulation did not affect NMA in the synchronization task. This finding is not consistent with the predictions based on the SAM. Discrepancies in sensitivity to stimulus intensity between sensorimotor synchronization and reaction-time tasks point to the involvement of different timing mechanisms in these two tasks.     

Dissociable perceptual-learning mechanisms revealed by diffusion-model analysis

Performance on perceptual tasks improves with practice. Most theories address only accuracy data and tacitly assume that perceptual learning is a monolithic phenomenon. The present study pioneers the use of response time distributions in perceptual learning research. The 27 observers practiced a visual motion-direction discrimination task with filtered-noise textures for four sessions with feedback. Session 5 tested whether the learning effects transferred to the orthogonal direction. The diffusion model (Ratcliff, Psychological Review, 85, 59–108, 1978) achieved good fits to the individual response time distributions from each session and identified two distinct learning mechanisms with markedly different specificities. A stimulus-specific increase in the drift-rate parameter indicated improved sensory input to the decision process, and a stimulus-general decrease in nondecision time variability suggested improved timing of the decision process onset relative to stimulus onset (which was preceded by a beep). A traditional d’ analysis would miss the latter effect, but the diffusion-model analysis identified it in the response time data.     

Inhibitory control and response selection in problem solving: how cotton-top tamarins (Saguinus oedipus) overcome a bias for selecting the larger quantity of food

When presented with a choice between 1 and 3 pieces of food in a type of reversed contingency task, 4 cotton-top tamarins (Saguinus oedipus) consistently chose the 3 pieces of food and received nothing, even though the choice of 1 piece would have yielded 3. However, in a task in which the tamarins received the 1 piece of food when they chose it, all subjects learned to select 1 over 3. Thus, the tamarins' prior failure on the reversed contingency task did not result entirely from an inherent inability to suppress the prepotent response of reaching to the larger of 2 quantities of food. After the experience of selecting the smaller quantity and receiving it, all of the tamarins solved the version of the reversed contingency task that they failed initially. These results suggest that the tamarins' initial failure may have reflected a difficulty with selecting an alternative response option.     

Working-memory load decreases mappings effects in spatial-compatibility tasks

The paper investigates the role of working memory for the translation of spatial stimuli into spatial responses. Performance is typically superior with compatible mapping between stimulus and response locations than with incompatible mapping (spatial-mapping effect). According to popular dual-route models, the spatial-mapping effect may arise from differences in the effectiveness of S–R translation, and/or from the effects of automatic response activation. Working-memory load should not affect the automatic route, but delay S–R translation in the incompatible condition, increasing the mapping effect under load. These predictions were tested in two dual-task experiments. Participants performed a spatial-compatibility task with compatible or incompatible S–R mapping, while memorizing different amounts of spatial (Experiment 1) or verbal (Experiment 2) information. In both experiments, working-memory load decreased the mapping effect, but did not eliminate it. Results are at odds with the dual-route account. An alternative explanation based on the response-discrimination account (Ansorge and Wühr in J Exp Psychol Hum Percept Perform 30:365–377, 2004) is proposed.     

Quantity judgments of sequentially presented food items by capuchin monkeys (Cebus apella)

Recent assessments have shown that capuchin monkeys, like chimpanzees and other Old World primate species, are sensitive to quantitative differences between sets of visible stimuli. In the present study, we examined capuchins’ performance in a more sophisticated quantity judgment task that required the ability to form representations of food quantities while viewing the quantities only one piece at a time. In three experiments, we presented monkeys with the choice between two sets of discrete homogeneous food items and allowed the monkeys to consume the set of their choice. In Experiments 1 and 2, monkeys compared an entirely visible food set to a second set, presented item-by-item into an opaque container. All monkeys exhibited high accuracy in choosing the larger set, even when the entirely visible set was presented last, preventing the use of one-to-one item correspondence to compare quantities. In Experiment 3, monkeys compared two sets that were each presented item-by-item into opaque containers, but at different rates to control for temporal cues. Some monkeys performed well in this experiment, though others exhibited near-chance performance, suggesting that this species’ ability to form representations of food quantities may be limited compared to previously tested species such as chimpanzees. Overall, these findings support the analog magnitude model of quantity representation as an explanation for capuchin monkeys’ quantification of sequentially presented food items.

Relative numerosity discrimination by chimpanzees (Pan troglodytes): evidence for approximate numerical representations

Two adult chimpanzees were trained on a relative “numerosity” discrimination task. In each trial, two arrays containing different numbers of red dots were presented on a CRT monitor. The subjects were required to choose the array containing the larger number of dots. In Experiment 1, using numerosities between 1 and 8, 28 different pairs were presented repeatedly, and accuracy scores were analyzed to explore which cues the chimpanzee subjects utilized to perform the task. Multiple regression analyses revealed that the subjects’ performance was (1) not simply controlled by the “numerical” difference between arrays, but that it was (2) best described by Fechner’s Law–that is accuracy increased linearly with the logarithmic value of the numerical difference between arrays divided by the number in the larger of the two arrays. This relationship was maintained when using much larger numerosities (Experiment 3). In Experiment 2, the chimpanzees were tested on the effects of total area and density by manipulating dot size and presentation area. The results revealed that these factors clearly affected the subjects’ performance but that they could not alone explain the results, suggesting that the chimpanzees did use relative numerosity difference as a discriminative cue.     

Unitization,similarity, and overt attention in categorization and exposure

Unitization, the creation of new stimulus features by the fusion of preexisting features, is one of the hypothesized processes of perceptual learning (Goldstone Annual Review of Psychology, 49:585–612, 1998). Some argue that unitization occurs to the extent that it is required for successful task performance (e.g., Shiffrin & Lightfoot, 1997), while others argue that unitization is largely independent of functionality (e.g., McLaren & Mackintosh Animal Learning & Behavior, 30:177–200, 2000). Across three experiments, employing supervised category learning and unsupervised exposure, we investigated three predictions of the McLaren and Mackintosh (Animal Learning & Behavior, 30:177–200, 2000) model: (1) Unitization is accompanied by an initial increase in the subjective similarity of stimuli sharing a unitized component; (2) unitization of a configuration occurs through exposure to its components, even when the task does not require it; (3) as unitization approaches completion, salience of the unitized component may be reduced. Our data supported these predictions. We also found that unitization is associated with increases in overt attention to the unitized component, as measured through eye tracking.     

Contrast class cues and performance facilitation in a hypothesis-testing task: evidence for an iterative counterfactual model

Hypothesis-testing performance on Wason’s (Quarterly Journal of Experimental Psychology 12:129–140, 1960) 2–4–6 task is typically poor, with only around 20% of participants announcing the to-be-discovered “ascending numbers” rule on their first attempt. Enhanced solution rates can, however, readily be observed with dual-goal (DG) task variants requiring the discovery of two complementary rules, one labeled “DAX” (the standard “ascending numbers” rule) and the other labeled “MED” (“any other number triples”). Two DG experiments are reported in which we manipulated the usefulness of a presented MED exemplar, where usefulness denotes cues that can establish a helpful “contrast class” that can stand in opposition to the presented 2–4–6 DAX exemplar. The usefulness of MED exemplars had a striking facilitatory effect on DAX rule discovery, which supports the importance of contrast-class information in hypothesis testing. A third experiment ruled out the possibility that the useful MED triple seeded the correct rule from the outset and obviated any need for hypothesis testing. We propose that an extension of Oaksford and Chater’s (European Journal of Cognitive Psychology 6:149–169, 1994) iterative counterfactual model can neatly capture the mechanisms by which DG facilitation arises.     

Focus on the essential: all great apes know when others are being attentive

When begging for food, all great ape species are sensitive to a human’s attention. However, studies investigating which cues are relevant for chimpanzees to assess the attentional state of others have produced highly inconsistent results. Some have suggested chimpanzees differentiate attention based on the status of the face or even the eyes, while others have indicated that body posture alone is the relevant cue. Kaminski et al. (Anim Cogn 7:216–223, 2004) compared the behaviour of chimpanzees, bonobos and orangutans while begging for food from a human experimenter who systematically varied his face and body orientation. Their results indicated that both factors, face and body orientation, affect apes’ begging behaviour. The authors claimed that while body orientation provides information about the experimenter’s general disposition to offer food, the visibility of the face provides information about the human’s attentional state. In the current study, we tested this hypothesis with all four great apes species. However, unlike Kaminski et al. (Anim Cogn 7:216–223, 2004), the experimenter was able to hand over food regardless of body orientation. The results show that as soon as the offering of the food was no longer restricted, the orientation of the face became the key factor. Therefore, we present the first evidence that all great ape species are able to assess the attentional state of a recipient based on the orientation of the face.     

Bayesian inference for the information gain model

One of the most popular paradigms to use for studying human reasoning involves the Wason card selection task. In this task, the participant is presented with four cards and a conditional rule (e.g., “If there is an A on one side of the card, there is always a 2 on the other side”). Participants are asked which cards should be turned to verify whether or not the rule holds. In this simple task, participants consistently provide answers that are incorrect according to formal logic. To account for these errors, several models have been proposed, one of the most prominent being the information gain model (Oaksford & Chater, Psychological Review, 101, 608–631, 1994). This model is based on the assumption that people independently select cards based on the expected information gain of turning a particular card. In this article, we present two estimation methods to fit the information gain model: a maximum likelihood procedure (programmed in R) and a Bayesian procedure (programmed in WinBUGS). We compare the two procedures and illustrate the flexibility of the Bayesian hierarchical procedure by applying it to data from a meta-analysis of the Wason task (Oaksford & Chater, Psychological Review, 101, 608–631, 1994). We also show that the goodness of fit of the information gain model can be assessed by inspecting the posterior predictives of the model. These Bayesian procedures make it easy to apply the information gain model to empirical data. Supplemental materials may be downloaded along with this article from .     

Informed inferences of unknown feature values in categorization

Many current computational models of object categorization either include no explicit provisions for dealing with incomplete stimulus information (e.g. Kruschke, Psychological Review 99:22–44, 1992) or take approaches that are at odds with evidence from other fields (e.g. Verguts, Ameel, & Storms, Memory & Cognition 32:379–389, 2004). In two experiments centered around the inverse base-rate effect, we demonstrate that people not only make highly informed inferences about the values of unknown features, but also subsequently use the inferred values to come to a categorization decision. The inferences appear to be based on immediately available information about the particular stimulus under consideration, as well as on higher-level inferences about the stimulus class as a whole. Implications for future modeling efforts are discussed.