PropertyValue
is nif:broaderContext of
nif:broaderContext
is schema:hasPart of
schema:isPartOf
nif:isString
  • Seven chimpanzees (six of whom also participated in Experiments 1b–3) participated in Experiment 1a (Table 1). The participants belonged to two social groups, containing a total of 12 individuals, which are housed at the Primate Research Institute, Kyoto University (KUPRI; Inuyama, Aichi, Japan). All participants were born in captivity except for Ai, who was brought to KUPRI from the wild when she was about one year old (details are available in the Great Ape Information Network, see Table 1). Their living environment includes an outdoor compound (700 m2) and attached indoor compounds [16]. From their living facilities, the chimpanzees have views of many houses in the city of Inuyama, from different angles. All chimpanzees had full access to food and water during the study. The chimpanzees had previous experience with cognitive tasks involving touch screens, including activities related to numerical sequence learning, short-term memory, acquisition of an artificial language, visual attention and visual search, and facial recognition [5, 33–36]. All procedures adhered to the Japanese Act on the Welfare and Management of Animals. The daily care and use of the chimpanzees adhered to the 2010 Guidelines for the Care and Use of Laboratory Primates of KUPRI. The research proposal was approved by the Animal Welfare and Animal Care Committee of KUPRI and also by the Animal Research Committee of Kyoto University. The participants sat in an experimental booth and performed tasks on a touch screen with a 15-inch LCD display (1,024 × 768 pixels, 1 pixel = 0.297 mm). Chimpanzees faced the screen head-on; screens were not tilted significantly in any direction (Fig 1). During the experiments, the chimpanzees could move freely, but they would always sit in front of the screen about 40 cm away. They always kept their postures natural and relaxed, and heads upright. When the chimpanzees made the correct choice in a trial, a piece of apple or raisin was provided via a feeder in conjunction with a chime sound. When they made a wrong choice, an error buzzer sounded, and no food was provided. The food reward was delivered by a universal feeder via a tube to a food tray placed at the bottom of the display. All experimental events were controlled by a computer. The experimental programs were written and operated with Microsoft® Visual Basic® 2010 software (Microsoft Corp.; Redmond, WA, USA). Each experiment contained two or more conditions (e.g. intact body, house, etc. ), and each condition was repeated for eight sessions. Each session included 40 trials: 20 were upright trials, in which all stimuli had the upright orientation, and the other 20 were inverted trials, in which all stimuli were inverted. The upright and inverted trials were mixed, and the order was randomly determined. Each condition included 40 pictures: 20 were in the upright position, and the remaining 20 were identical to the first 20 but in the inverted position (i.e. 40 pictures in total). All pictures were in black and white, and balanced in terms of luminance. Pictures of bodies and houses were about 400 * 400 px. Face stimuli presented in Experiment 3 were about 80 (width) * 115 (height) px, similar to faces presented in other conditions. In each trial, one stimulus served as the sample, and another stimulus (Experiment 1a) or three other stimuli (Experiments 1b, 2a, 2b, 3) was/were randomly chosen from those with the same orientation. The correct stimuli and their locations were counterbalanced in the experiment. The original pictures were manipulated using Adobe Photoshop software (Adobe Systems; San Jose, CA, USA) and Pixelmator (Pixelmator Team Ltd.; Vilnius, Lithuania). Some pictures were obtained from the Internet, and the rest were provided by Kumamoto Sanctuary, Wildlife Research Center of Kyoto University. We evaluated the similarity of all the pictures (more than the desired number) before choosing those used in the experiment. The participants were not familiar with the specific chimpanzees or houses in the stimulus pictures. Both Experiment 1a and Experiment 1b had two conditions: 1) intact body and 2) house (Fig 3A). The house condition was a control condition [3, 5, 21]. In Experiment 2a, there were three conditions: 1) the intact-body condition, which was used as the control; 2) the only-body-clear condition, in which the face was replaced by a mosaic pattern; and 3) the only-face-clear condition, in which the body, with the exception of the face, was replaced by a mosaic pattern (Fig 3B). In Experiment 2b, there were four conditions: 1) the intact-body condition, which was used as a control; 2) the only-body condition, in which there were no faces; 3) the only-face condition, in which only the faces were shown; and 4) the body-silhouette condition, in which intact chimpanzee bodies were solid black (Fig 3C). In Experiment 3, there were two conditions: 1) the face-without-contour condition and 2) the face-silhouette condition (Fig 3D). In the face-without-contour condition, the outline of the face was replaced by an oval shape of 75 (width) * 110 (height) px, and only the face contents remained. Under the face-silhouette condition, the original contours remained but were filled with solid black. A zero-delayed matching-to-sample task (Fig 2) was used. Experiment 1a offered two choices per trial; the other experiments offered four choices per trial. In each trial, the start key (a circle) initially appeared at the bottom centre of the touch screen against a white background. After the chimpanzees touched the start key, the sample stimulus appeared at the centre of the screen. The sample remained for 500 ms, then disappeared upon touching. At the same time, image choices appeared side by side in the top part of the screen. A correct choice was one in which the chimpanzees chose the image that was the same as the sample. Correct choices were followed by a piece of food as a reward accompanied by a chime sound. Otherwise, a buzzer would sound and no food would be presented. The inter-trial-interval (ITI) was 1.5 s, and the timeout was 2 s. Accuracy and response-time data were recorded and analysed. We compared data to determine whether the error rate (100%—accuracy) and the response times differed between all upright trials and all inverted trials under each condition. According to previous studies (e.g. [6]), if either the error rate or the response time were higher in the inverted trials than in the upright trials, an inversion effect would be deemed to be present. Data were analysed using a generalised linear mixed model (GLMM) in R 3.3.1 [37] using the lme4 package [38]. We calculated the error rates of the 20 upright trials and the 20 inverted trials on a session-by-session basis. The distribution of the error rate was binomial. The fixed effect was orientation (upright or inverted). The random effects were participant ID and session number. Response time refers to the duration of time from the chimpanzee touching the sample to when s/he chose any one of the alternatives. Only the response times from correct trials were included in the analyses. We calculated the average response time of all correct upright trials and all correct inverted trials on a session-by-session basis. During the experiment, unexpected sounds from outside the experimental booths were sometimes heard, and the chimpanzees were interrupted; therefore, we initially discarded response times that were longer than the mean value plus 3 SDs. We then calculated the average of the remaining data points and used them for further analyses. The response-time data were normally distributed. We compared the null model and the model with the fixed-effect orientation via analysis of variance (ANOVA) using the anova() function to examine whether there were differences in the response times for the two orientations. Both models included the random effects of participant ID and session number.
rdf:type