Observing real-time social interaction via telecommunication methods in budgerigars (Melopsittacus undulatus)

Behavioural Processes 128 (2016) 29–36

Contents lists available at ScienceDirect

Behavioural Processes journal homepage: www.elsevier.com/locate/behavproc

Observing real-time social interaction via telecommunication methods in budgerigars (Melopsittacus undulatus) Yuko Ikkatai a,b , Kazuo Okanoya a,c , Yoshimasa Seki b,c,∗ a b c

Brain Science Institute, RIKEN, Wako, Japan Faculty of Letters, Aichi University, Toyohashi, Japan Graduate School of Arts and Sciences, The University of Tokyo, Tokyo, Japan

a r t i c l e

i n f o

Article history: Received 30 October 2015 Received in revised form 28 March 2016 Accepted 30 March 2016 Available online 31 March 2016 Keywords: Budgerigar Melopsittacus undulatus Synchrony Telecommunication Vocal production

a b s t r a c t Humans communicate with one another not only face-to-face but also via modern telecommunication methods such as television and video conferencing. We readily detect the difference between people actively communicating with us and people merely acting via a broadcasting system. We developed an animal model of this novel communication method seen in humans to determine whether animals also make this distinction. We built a system for two animals to interact via audio-visual equipment in real-time, to compare behavioral differences between two conditions, an “interactive two-way condition” and a “noninteractive (one-way) condition.” We measured birds’ responses to stimuli which appeared in these two conditions. We used budgerigars, which are small, gregarious birds, and found that the frequency of vocal interaction with other individuals did not differ between the two conditions. However, body synchrony between the two birds was observed more often in the interactive condition, suggesting budgerigars recognized the difference between these interactive and non-interactive conditions on some level. © 2016 Elsevier B.V. All rights reserved.

1. Introduction Humans communicate with one another not only face-to-face but also via modern telecommunication methods. “Skype” and “FaceTime” are well-known tools and some people are starting to customize these novel applications for communication between pets and their owners, and even between animals. For example, owners can see and “talk” with their dogs even while they are not at home using iCPooch—Internet Pet Treat Dispenser (ICPOOCH01US, iCPooch). Some studies have explored owner-pet video-chat systems (e.g. Golbeck and Neustaedter, 2012; Neustaedter and Golbeck, 2013; Murata et al., 2014). Most of these attempts were undertaken as part of engineering or animal welfare studies. Additionally, we found several video clips uploaded on YouTube demonstrating that pet owners tried to contact their pets via “Skype” and “FaceTime”. When contacting their pets via these telecommunication tools, pet owners seem to assume that their pets recognize that they are communicating with their owners in real-time, and are not merely watching a video image. However, no

∗ Corresponding author at: Faculty of Letters, Aichi University, Machihata-machi 1-1, Toyohashi, 441-8522, Japan. E-mail address: [email protected] (Y. Seki). http://dx.doi.org/10.1016/j.beproc.2016.03.020 0376-6357/© 2016 Elsevier B.V. All rights reserved.

psychophysical studies have evaluated whether this assumption is correct or not. Therefore, the present study attempted to clarify this issue qualitatively and quantitatively. Establishing an animal model for studying this modern communication style will help us understand the nature of communication in addition to creating a new paradigm of animal experiments. In addition, it may contribute to studies of the “Theory of Mind (ToM)” in animals. In humans, to establish telecommunication between “person A” and “person B”, it is necessary for “person A” to confirm that he can properly receive information from “person B”. In addition, “person A” must also confirm that “person B” can receive the information that “person A” is transmitting. Thus, telecommunication requires an understanding of the viewpoint of others, which is integral to ToM. Previous studies have reported that animals can successfully discriminate others’ faces when presented visually on a monitor (e.g. budgerigars, Melopsittacus undulatus, Brown and Dooling, 1992; chimpanzees, Pan troglodytes, Parr et al., 2000; dogs, Canis familiaris, Racca et al., 2010). Patton et al. (2010) used artificial static images of female pigeons to investigate courtship behavior of males (Columba livia) and they showed that local features of the face were important to these courtship behaviors. Pigeons have been shown to discriminate real-time self-video images from delayed self-video images (Toda and Watanabe, 2008). Thus, animals are

30

Y. Ikkatai et al. / Behavioural Processes 128 (2016) 29–36

Table 1 Classification and definition of 10 behaviors. Behavior

Definition

Drinking Feeding Jumping to the monitor Walking to the monitor Yawning Preening Leg-stretching Wing-stretching Scratching Ruffling

Takes water into the beak and swallows Takes seeds Holds onto the cage wall located between the perch and monitor Approaches the cage floor located between the perch and monitor Opens beak and deeply inhales Nibbles own feathers Extends limbs Extends flight feathers Rubs own body using legs Erects whole body feathers no longer than 1–2 s

Fig. 1. Schematic of the experimental setup. A bird was put in a metal-mesh wire cage placed inside the chamber. A monitor with a built-in-loudspeaker, a microphone and a micro CCD camera (a) were located in front of the cage for the bird to be able to interact with another bird. Audio-visual equipment was controlled by 3 PCs (b).

likely sensitive to both static and video images of other individuals (review in D’Eath, 1998; Fleishman and Endler, 2000). Moreover, such video images could modify animal behavior. Non-human primates were shown to yawn more often after viewing video playback of yawning in other conspecifics (chimpanzees, Anderson et al., 2004; stumptail macaques, Macaca arctoides, Paukner and Anderson, 2006). Courtship behaviors in birds have been examined using video images (pigeons, Shimizu, 1998; bengalese finches, Lonchura striata var. domestica, Takahashi et al., 2005, Takahashi and Okanoya, 2013; zebra-finches, Taeniopygia guttata castanotis, Ikebuchi and Okanoya, 1999). In addition, courtship in spiders have been studied in this way (jumping spiders, Maevia inclemens, Clark and Uetz, 1992; wolf spiders, Schizocosa ocreata, Uetz and Smith 1999), as well as fish (three-spined sticklebacks, Gasterosteus aculeatus, Rowland et al., 1995; guppies, Poecil reticulate, KodricBrown and Nicoletto, 1997). Video images also induce foraging (manakins, L. punctulata, Rieucau and Giraldeau, 2009) and imitative behaviors (budgerigars, Mottley and Heyes, 2003; Mui et al., 2008). Additionally, Watanabe and Troje (2006) demonstrated that computer-generated (CG) pigeons could serve as experimental visual stimuli. Pigeons discriminated between CG pigeons displaying normal movements from those displaying movements that were physically impossible. However, there are few studies examining whether animals can recognize another animal presented in both a video and in person as being the same individual. For example, female Japanese quail (Coturnix japonica) showed a preference for males that had been presented through a monitor previously (Ophir and Galef, 2003). Rooks (Corvus frugilegus) were able to recognize affiliated conspecifics from unaffiliated conspecifics even though they were shown as video images (Bird and Emery, 2008). These studies indicate that birds can make a connection between an individual presented as a video image and an individual viewed in person. The responses to video-images can be different

depending on the type of the video playback (review in King, 2015). For example, jacky dragons (Amphibolurus muricatus) showed more aggressive displays with playback of video-images of a submissive individual, whereas video-images of an aggressive dragon inhibited such aggressive displays (Ord and Evans, 2002). Video images of conspecific animals can elicit spontaneous operant responses in primates (bonnet macaques, Macaca radiate, Swartz and Rosenblum, 1980; Andrews and Rosenblum, 1993; chimpanzees, Fujita and Matsuzawa, 1986; Japanese macaques, Macaca fuscata, Tsuchida and Izumi, 2009). Such “sensory reinforcement” has been also reported in birds. Gilbertson (1975) showed that a direct visual contact with a real bird functioned as a reinforcement in pigeons (although Linton (1981) failed to replicate these results). Static images of conspecifics (European starlings, Sturnus vulgaris, Perret et al., 2015) and conspecific song (zebra finches, Adret, 1997) also could function as a primary reinforcer. Therefore, we posed the question: how do birds behave when we provide them with a real-time interactive environment via telecommunication tools? Previous studies showed that audio-visual presentation of stimuli could elicit some behavioral responses in budgerigars, a small parrot species that establish flocks (Wyndham, 1980). Such studies have demonstrated stronger vocal production in response to playback of their mates’ calls compared with the calls of others (Ali et al., 1993), behavioral contagion with viewing video playback of others (Mui et al., 2008), and contagious yawning with playback of yawning of others (Gallup et al., 2015). These studies suggest that this species is fairly suitable for the present study rather than rodents, which require olfactory and somatosensory inputs for recognition of other individuals. Thus, we built a system for two budgerigars to interact via audio-visual equipment between two separate cages. We observed birds’ behavior across three experimental conditions: (1) realtime, two-way communication (Paired condition), (2) “one-way”

Y. Ikkatai et al. / Behavioural Processes 128 (2016) 29–36

communication in which one bird viewed a live video of the other bird (Pseudo-paired condition), and (3) one bird viewed an empty cage (Solo condition). We expected that interactive behavior of the birds would be observed more often in the Paired condition than in the Pseudo-paired and Solo conditions. 2. Materials and methods 2.1. Animals Eight budgerigars (four males, and four females, ranging in age from 2 to 4 years old) participated in this study. We obtained the birds from local suppliers when they were sexually mature. We began the experiment in four birds. Then, we added four more birds came from another local supplier to reinforce the robustness of the experimental results. The birds were therefore tested as two different groups. Both groups consisted of four birds (two males and two females in each group). Within each group, same-sex couples of birds were paired together and kept in one cage (W 23.0 cm × L 28.5 cm × H 31.0 cm). Seed mixture, vitamins, calcium and water were available ad libitum. The cages were placed at an animal rearing room of our laboratory in RIKEN-BSI during the experimental period and the room was maintained at 24 ± 4 ◦ C on a 12:12-h lightdark cycle with light on at 0700. All experimental procedures were approved by the RIKEN animal experiments committee and they complied with RIKEN BSI guidelines. 2.2. Apparatus We built an online communication system for two birds to interact in real time (Fig. 1a). Audio-visual equipment was placed in two sound-attenuated chambers (inside dimensions: W 58.9 cm × L 40.9 cm × H 37.0 cm). Equipment included a TFT monitor (frame size: 800 × 600) with a built-in loudspeaker (LCD-8CS, PLANEX), a microphone (PRO35, Audio-technica), a micro CCD camera (WAT23OA, Watec) and a metal wire cage (W 30.0 cm × L 15.5 cm × H 22.0 cm) (Fig. 1a). The camera and the microphone were used for sending audio-visual signals to the other chamber. The visual signals from each camera were separately inputted to personal computers via USB video capture devices (PC-SDVD/U2G2, frame size: 720 × 480) (Fig. 1b). All video signals were recorded as noncompressed AVI files (frame rate: 29.97 fps, frame size: 720 × 480). All vocalizations were recorded as WAV files using independent multi-channel sound recording software (Avisoft-RECORDER, version 5.2; Avisoft Bioacoustics, Germany) at a sampling frequency of 16 kHz. The cage was equipped with a food cup, a water cup and a perch. A LED was attached on the side wall of each chamber. The LED faced the camera, and could be triggered from outside by the experimenter, to signal the beginning of the session on the movie files. The sound pressure level from the loud speaker was set at 45 dB at the position of the birds’ head using a playback of 3-kHz pure tone at 70 dB SPL in another chamber. The sound level of the stimuli was adjusted to be relatively low in amplitude to prevent a feedback loop in which one bird’s call is detected and transmitted back and forth between the cages. Because the frequency range of a contact call is at around 2–4 kHz (Dooling, 1986), a 3-kHz pure tone was appropriate for this calibration. The background noise of the two chambers was 33.1 ± 0.85 dB (mean ± SD). 2.3. Procedures We conducted three experimental conditions (Solo, Paired, and Pseudo-paired; 65-minute-recording for each session; 112 h in total). In the Solo condition, we put a single bird either in cage 1 or 2

31

only (Fig. 1b) so that the monitor displayed an empty cage throughout the session. In the Paired condition, each cage held a single bird and those two birds could interact online in real-time throughout the session. In the Pseudo-paired condition, two birds were used as in the Paired condition; however, “interaction” could not be established between the birds. Bird A could see and hear the Bird B in real time, but Bird B saw and heard a previously recorded movie of Bird A (frame size: 720 × 480). Therefore, it was impossible for the two birds to mutually interact in real time. All birds within each group experienced the three conditions. For the Paired and Pseudo-paired conditions, we tested all combinations of two dyads within the group. The sessions were performed twice in each dyad with an alternation of the chambers. We conducted the Solo condition for 2 sessions with alternation of the chambers, the Paired condition for 6 sessions (3 dyadic combinations × alternation of the chambers), and the Pseudo-paired condition for 6 sessions (3 dyadic combinations × alternation of the chambers) per bird. In some analyses, we randomly chose two video-recordings from all of the Solo recordings (not empty side) without duplication to make “simulative dyads” for the Solo conditions (see details below). All sessions were conducted from 0900 to 1700 between July and September 2013. The experimental order was counterbalanced across groups and randomized within each condition. The behavior of two birds were independently recorded in each single AVI file; however the beginning of each experimental session was signalled by LEDs which were triggered by the experimenter and the signal was recorded on both AVI files so as to evaluate the synchrony of the subjects’ behavior. 2.4. Data analysis The AVI files were used to measure locomotive activity, locomotive synchrony and to visually inspect the behavior of the subjects. The WAV files were used to analyze timing and quantity of the vocalizations of the subjects. Statistical tests were conducted using R 2.13.0 (R foundation for Statistical Computing). Alpha was set at 0.05. 2.4.1. Locomotive activity Because budgerigars are a highly social species, locomotive activity might be facilitated when birds react to one another’s behavior even if this is through telecommunication rather than inperson interaction. Therefore, we hypothesized that the locomotive activity would be higher in the Paired condition than in the other conditions. For each AVI file, we extracted 3601 video frames from the beginning of the video frame in which the LED was triggered, in one-second intervals. The frames were transformed into grayscale pictures. Each picture was represented as a two dimensional matrix (each single frame with 480 lines (x-axis) and 720 pixels (y-axis), and each cell had the intensity information of the black-and-white pixels). We transformed each element of the matrix into a z-score: {(each individual element—the mean of the matrix elements)/(the standard deviation)}. We subtracted each z-score of one matrix from the corresponding z-score of the next matrix, and took the absolute value. Then, we averaged all differences between the two matrices as a “difference score” of those two matrices. This results in a quantitative evaluation of the movement of the objects (i.e., the birds) in the video, frame by frame. This calculation was executed over the 3601 video frames to give a sequence of scores, termed the score sequence. Video frames included some background noises (such as high-frequency fluctuations of florescent lighting, highfrequency flickers with digitizing analog signals to image pixels), thus, many small jags which reflected those noises appeared on the score sequences. So, we applied a Butterworth low-pass filter

32

Y. Ikkatai et al. / Behavioural Processes 128 (2016) 29–36

(second-order), a function provided by Matlab (version R2012b; The MathWorks, USA), to remove such high-frequency jags from the score sequences and to smooth them. Because fine adjustment of the system at the beginning of each session might have disturbed the birds, the first 300 frames (5 min of data) were considered as an adaption period and excluded from further analyses. The values were summed in each session and averaged as a “locomotive activity” by condition in each bird. Then, it was compared among the Solo, Paired and Pseudo-paired conditions using a Friedman test.

2.4.2. Locomotive synchrony In humans, locomotive synchrony is often observed when two people are talking face-to-face (Paxton and Dale, 2013). Thus, we hypothesized that budgerigars would synchronize their movements with others more in the Paired condition than in the Pseudo-paired condition. We measured time-lagged cross-correlation coefficients (30second time window) to evaluate the synchrony of the birds’ locomotive activity. A higher value at a lag 0 means the two birds of the dyad were synchronized in their movements (i.e., after one bird moved, the other bird moved within the corresponding time window). A higher value at a lag +1 means the movements of the two birds were separated in time such that one bird within the dyad moved mostly within the next time window. Two score sequences of locomotive activity, which were calculated as described in Section 2.4.1, for each dyad were applied for the Paired and Pseudo-paired conditions. We then calculated a crosscorrelation coefficient between two score sequences. This method was used in Paxton and Dale (2013). For each Solo condition experiment, we made a “simulative dyad” as described in Section 2.3. The cross-correlation coefficients (r) were pooled and averaged by condition in each dyad. The coefficients were compared among the Solo, Paired, and Pseudo-paired conditions using a Friedman test.

2.4.3. Classification of behavior To analyze, not merely locomotive activity but, appearance frequency and it’s timing of each behavioral class, we visually inspected the AVI files (112 h in total) and created ethograms with 10 behaviors (Table 1) that are typically observed in budgerigars (e.g., Brockway, 1964a, 1964b; Trillmich 1976). The experimenter (Y.I.) manually counted the number of times that each behavior occurred. The specific behavior, the time point, and the actor of the behavior were described for each session. The behavior counts were pooled and averaged by condition in each bird. We compared the behavior counts among the Solo, Paired and Pseudo-paired conditions using a Friedman test.

2.4.4. Behavioral contagion We explored locomotive synchrony between two birds as described in Section 2.4.2. However, it is still unknown what specific types of behaviors tend to be synchronized between birds. We therefore compared the number of behavioral contagion between the Paired and Pseudo-paired conditions, examining ten individual behaviors (see Table 1). We calculated the latency between the onset of a “Behavior A” in one bird and the onset of “Behavior A” in the other. Following the methods of Palagi et al. (2009) and Gallup et al. (2015), only the behaviors which had a latency of less than 5 min were collected and considered as behaviorally contagious. The number was pooled in each bird. For each Solo condition experiment, we made a “simulative dyad” as described in Section 2.3. We applied a Friedman test to assess the number of contagious behaviors among the Paired, Pseudo-paired and Solo conditions.

Fig. 2. A spectrographic example of vocal exchange. A contact-call exchange was defined when the inter call interval (ICI) between two birds was 1 s or less. The X axis represents time.

2.4.5. Vocal production In addition to locomotive activity, vocal production may also be facilitated when birds react to one another’s vocalizations through telecommunication methods. We hypothesized that vocal production would increase more in the Paired than in the Pseudo-paired condition. To test this, we manually counted the number of bouts of vocal production by means of visual inspection of the sound spectrograms (FFT length = 256, overlap of the time window = 75%). A bout was defined as a successive vocal output in which silent intervals were less than 10 s in duration. The number of bouts produced in each condition was pooled and averaged for each bird. The number was compared among three conditions using a Friedman test. In addition, we manually counted the number of contact-call exchanges between two birds. It is well known that when a bird is separated from group members, the bird often emits “contact calls”. A contact-call exchange was defined as the case in which the inter-call interval (ICI) between two contact calls emitted by two different budgerigars was 1 s or less (Fig. 2). A “contact call” in budgerigars is a single vocal sound around 150-msec in duration (Dooling, 1986) and each individual has one or more contact call types (Farabaugh et al., 1994). Such vocal exchanges may facilitate a real-time response from other birds even if the original call was given via audio-visual presentation and not in person. For each Solo condition experiment, we made a “simulative dyad” as described in Section 2.3. The number of vocal exchanges in the three conditions (Paired, Pseudo-paired and Solo) was pooled and averaged for each dyad. The differences of the number of exchanges were compared among the Paired, Pseudo-paired and Solo conditions using a Friedman test. We recorded vocal sounds of the two birds using two microphones, which have cardioid directivity in two independent channels. This enabled us to avoid potential contamination of the contact calls of two birds. Even if one microphone recorded both birds’ calls, we could easily differentiate the calls as the sound intensities were extremely different. 3. Results 3.1. Locomotive activity The locomotive activity of the subjects was greater both in the Paired and Pseudo-paired conditions compared with the Solo con-

Y. Ikkatai et al. / Behavioural Processes 128 (2016) 29–36

33

Fig. 3. Locomotive activity. Thick horizontal lines show medians, boxes show quartiles, and vertical dashed lines represent the maximum and minimum values. Significant differences were found between the Solo and Paired, and the Solo and Pseudo-paired conditions (Wilcoxon test with Bonferroni correction,*p < 0.05).

dition. The difference was statistically significant among conditions (Friedman test, ␹2 (2) = 12.25, p < 0.002, Fig. 3). Post-hoc tests using a Wilcoxon signed-rank test with Bonferroni correction revealed that activity level in the Paired condition was significantly higher than in the Solo condition (r = 0.94, p = 0.043) and that activity in the Pseudo-paired condition was higher than in the Solo condition (r = 0.94, p = 0.043). No significant difference was found between the Paired and Pseudo-paired conditions (r = 0.51, p = 0.424).

3.2. Locomotive synchrony We counted the number of sessions in which the crosscorrelation coefficients (r) of locomotive activity between the dyads were largest at lag 0, which means that the movements of the two birds were synchronous, and occurred one after the other (i.e. within a 30-second time window). This was observed in 7 sessions in the Paired, 2 sessions in the Pseudo-paired and 5 sessions in the Solo conditions. Chi-squired test revealed no significant difference for the factor of conditions (Friedman test, ␹2 (2) = 2.7, p = 0.257). However, the size of cross correlation coefficients at lag 0 was significantly different for the factor of the condition (␹2 (2) = 14, p < 0.001, Fig. 4). Post-hoc tests using a Wilcoxon signed-rank test with Bonferroni correction revealed that the Paired was larger than the Solo (r = 0.95, p < 0.001), the Pseudo-paired was larger than the Solo (r = 0.92, p = 0.013), and the Paired was larger than the Pseudopaired (r = 0.75, p = 0.040).

Fig. 4. Cross correlations coefficients at lag 0. Thick horizontal lines show medians, boxes show quartiles, and vertical dashed lines represent the maximum and minimum values. The size of the coefficient was significantly different across conditions (Wilcoxon test with Bonferroni correction,*p < 0.05).

3.4. Behavioral contagion 3.3. Classification and quantification of behavioral patterns The behavioral patterns differed depending on the condition (Fig. 5). As shown in Table 2, a Friedman test and post-hoc tests using a Wilcoxon signed-rank test with Bonferroni correction revealed that the mean frequencies of ruffling and wing-stretching were significantly higher in the two dyadic conditions (Paired and Pseudo-paired) than in the Solo condition.

Contagious behaviors were less observed in the Solo condition (Table 3). We observed some contagious behaviors and chose to analyze ruffling, yawning and wing-stretching because the number of observations of these behaviors was sufficient for statistical tests (Fig. 6). The number of mean contagious behaviors in the Paired condition was greater than that in the Pseudo-paired condition for ruffling, and wing-stretching (Fig. 6). However, the difference was not statistically significant (Table 3), suggesting that interactive

34

Y. Ikkatai et al. / Behavioural Processes 128 (2016) 29–36

Fig. 5. Mean frequencies of 10 common behaviors in budgerigars. Statistical significance determined by Friedman tests. Error bars indicate ± 1 SEM (*p < 0.05). Table 2 Differences in the frequency of behaviors among conditions. Behavior

Total observed number

Ruffling Wing-stretching Scratching Drinking Preening Walking to the monitor Feeding Leg-stretching Jumping to the monitor Yawning *

Friedman test

Paired

Pseudo-paired

Solo

␹2 (2)

393 177 50 67 67 17 100 31 166 152

351 156 44 74 76 31 81 25 101 164

29 7 0 6 6 0 6 0 7 45

12.25 12.97 10.30 9.85 8.32 6.00 5.79 9.00 3.90 3.16

Post-hoc test Paired vs. Solo

Paired vs. Pseudo

Pseudo vs. Solo

*

*

*

*

*

*

ns ns ns ns – – – –

ns ns ns ns ns ns – – – –

* * *

ns ns ns ns

*

ns ns ns ns – – – –

p < 0.05.

Table 3 Number of contagious behaviors. Behavioural classes

Ruffling Wing-stretching Yawning Jumping to the monitor Preening Feeding Leg-stretching Scratching Drinking Walking to the monitor *

Total observed number Paired

Pseudo-paired

Solo

171 65 30 22 8 8 5 4 2 0

152 50 30 15 12 1 2 2 2 0

4 0 15 0 0 0 0 0 0 0

Total

Friedman test

327 115 75 37 20 9 7 6 4 0

13.00 11.86 9.21 – – – – – – –

Post-hoc test

␹2 (2)

Paired vs. Solo

Paired vs. Pseudo

Pseudo vs. Solo

*

*

*

*

ns ns – – – – – – –

ns ns ns – – – – – – –

*

– – – – – – –

ns ns – – – – – – –

p < 0.05.

responses between birds did not facilitate contagious behaviors in dyads. 3.5. Vocal production A significant difference was found in the number of vocal bouts produced between conditions (mean ± SEM, Paired = 18.46 ± 4.26, Pseudo-paired = 20.40 ± 5.38, Solo = 3.38 ± 1.68, Friedman test; ␹2 (7) = 16.63, p = 0.020). Post-hoc tests using a Wilcoxon signed-rank test with Bonferroni correction revealed that significantly more vocal bouts were produced in the Paired than in the Solo condition (r = 0.94, p = 0.043). However, no significant difference was found between the Paired and Pseudo-paired (r = 0.17, p = 1.00) or between the Pseudo-paired and Solo conditions (r = 0.74, p = 0.13). This suggests that vocal production is likely to increase more in real-time communicative condition compared with Solo condition.

The mean number of vocal exchanges was greater in the Paired condition than in the Pseudo-paired and Solo conditions. However, again the difference was not significant (mean ± SEM, Paired = 2.50 ± 1.55, Pseudo-paired = 0.50 ± 0.32, Solo = 0.00 ± 0.00; Friedman test; ␹2 (2) = 5.76, p = 0.056), suggesting that the vocal exchange between birds in the communicative condition was not different from that of the video-image condition.

4. Discussion In this study, we qualitatively and quantitatively evaluated social interaction via telecommunication methods in budgerigars. As far as we know, this is the first study that has investigated social communication via telecommunication in non-human animals from the perspective of experimental psychology. It was very

Y. Ikkatai et al. / Behavioural Processes 128 (2016) 29–36

35

Fig. 6. Mean frequencies of 10 contagious behaviors in budgerigars. Friedman tests were conducted for ruffling, wing-stretching and yawning. Error bars indicate ± 1 SEM (*p < 0.05).

clear that the Paired and Pseudo-paired conditions resulted in the budgerigars becoming more locomotive than in the Solo condition, showing that presentation of audio-visual stimuli evoked a response from the subjects. This result is consistent with the previous studies described in the Section 1. In addition, we found that locomotive synchrony in the Paired condition was higher than that in the Pseudo-paired condition, suggesting that the birds recognized some differences between those two conditions. Human-human dyads that showed high preference for one another also showed higher synchrony of body movements (Paxton and Dale, 2013), thus, synchrony may be important for establishing preferred relationships (Tschacher et al., 2014). Assuming locomotive synchrony in budgerigars function as in humans, therefore, our results suggest that the birds might prefer the Paired condition more than both the Pseudo-paired and Solo conditions. However, from the statistical analyses, locomotive activity and vocal exchange were not significantly different between the Paired and Pseudo-paired conditions. This raises the possibility that whether the bird presented on the monitor is interactive or not may not matter to the birds in terms of how behavior is affected. However, we have two other possible interpretations as to why we detected the difference only in a few analyses; first, the intensity or amount of the stimulus may not be enough to induce the birds’ responses in the Paired condition. In a video playback study in Jacky dragons, the experimenter presented different video clips of dragons, depending on the dragon’s previous response. Behavioral responses were greater in this interactive situation than in a more conventional playback situation in which the same video clips were played back regardless of response (Ord and Evans, 2002). However, the birds’ behavior was not controlled in our study, so the stimulus intensity was dependent only on the activity of the birds themselves. Second, social deprivation may not be enough for the birds. Budgerigars could interact with their cage mates in their home cage after each session. Because birds had this social interaction available to them, the bird displayed on the audio-visual equipment might serve as a poor stimulus, and may not be enough to drive any active communication. We had expected that the activity level in the Paired condition would be quite larger than in the Pseudo-paired condition, because a feedback loop, or chain reaction, could occur in the Paired condition. However, the difference between the Paired and Pseudopaired conditions is likely to be modest. The results showed that subjects became more active when the other bird appeared on the monitor, suggesting that the birds recognized the stimulus as a significant signal but we did not find any evidence that the mutual

“interaction” affected the amount of the locomotion. One possible interpretation is that the live bird (Bird A) may have shown a high level of activity in the Pseudo-paired condition to compete for the attention of the other bird (Bird B) presented on a monitor. If Bird A interpreted Bird B’s behavior as engaging in some form of communication with another bird, Bird A may have moved more to attract the attention of Bird B. This could result in the same level of locomotive activity both in the Paired and Pseudo-paired conditions, but with less synchrony in the Pseudo-paired condition, which is what we observed. Another possible interpretation is that the birds preferred the Pseudo-paired condition much more than we had thought, even though the subjects recognized the “interaction” was only one-way. It is quite possible that some birds recognized that the individual that appeared on the monitor was merely a 2D image in the Paired condition. In this case, it should seem strange and unexpected for the focal bird that the stimulus birds returned some reaction in response to their behavior. Such situations might inhibit activity. In this kind of studies, we can also consider the effects of individual differences. The fact that the peak of cross correlation coefficients (r) varied depending on sessions might be a reflection of such individual differences. The dyads we used in this study were same-sex cage-mates or different-sex non cage-mates. However, if we used more affiliated dyads such as male-female pairs, we would expect their mutual responses to be stronger than what we observed in this study. Additionally, we should be cautious on several points. The first point is the sample size. The number of subjects was not so large due to the time consuming analyses (i.e. quantification of each behavioral class by means of visual inspection of video recordings). If we use some more subjects, the results might be clearer, although we used 2 groups consisted of 4 birds each (8 birds in total), so that the reliability of the results is likely to be confirmed. The second point is the method of setting the sound pressure level. To prevent a feedback loop between the audio devices, the sound pressure level from the loud speaker was set relatively low (see Section 2). The birds therefore saw another bird on a monitor in close proximity, but the corresponding vocalizations may have been perceived as coming from a distance. This mismatch between visual and auditory information might have inhibited vocal production. The third point is the small number of vocal exchanges (i.e., contact calls). Although the difference was not significant, the mean number of vocal exchanges in the Paired condition was five times that of the Pseudo-paired condition. Vocal interaction may therefore have been inhibited more in the Pseudo-paired condition than in the Paired condition. We may not have seen a significant

36

Y. Ikkatai et al. / Behavioural Processes 128 (2016) 29–36

difference between these conditions because of the low number of calls. Call convergence has been reported among male groups (Farabaugh et al., 1994), female groups (e.g. Hile and Striedter 2000) and male-female pairs (Hile et al., 2000) kept together. This call convergence should be one of the most interesting behaviors of budgerigars. Thus, we had expected the number of contact calls could be applied as an index to quantify their communication even in our research. However, the number of contact calls was so small in the experiments. Contact calls, as a function, are often emitted while a bird is separated from other birds. In the Paired and the Pseudo-paired conditions, the subjects always faced to a bird via the monitor. This situation may be not suitable for vocal exchange, which is a possible reason to explain the results. In conclusion, we found that locomotive synchrony was higher in a real-time, two-way interactive condition than in a one-way non-interactive condition. However, such social “communication” was likely limited, somewhat due to the audio-visual devices. It is therefore difficult to conclude that this kind of communication style is a complete substitute for direct, in-person communication between budgerigars, even if it could serve as a useful experimental paradigm. Acknowledgements This work was supported by JSPS KAKENHI Grant Number 24653210 to Y. S. References Adret, P., 1997. Discrimination of video images by zebra finches (Taeniopygia guttata): direct evidence from song performance. J. Comp. Psychol. 111, 115–125. Ali, N.J., Farabaugh, S., Dooling, R., 1993. Recognition of contact calls by the budgerigar (Melopsittacus undulatus). Bull. Psychonomic Soc. 31, 468–470. Anderson, J.R., Myowa-Yamakoshi, M., Matsuzawa, T., 2004. Contagious yawning in chimpanzees. Proc. R. Soc. Lond. B Biol. 271, 468–470. Andrews, M.W., Rosenblum, L.A., 1993. Live-social-video reward maintains joystick task performance in bonnet macaques. Percept. Motor Skill 77, 755–763. Bird, C.D., Emery, N.J., 2008. Using video playback to investigate the social preferences of rooks, Corvus frugilegus. Anim. Behav. 76, 679–687. Brockway, B.F., 1964a. Ethological studies of the budgerigar: reproductive behavior. Behaviour 23, 294–323. Brockway, B.F., 1964b. Ethological studies of the budgerigar: non-reproductive behavior. Behaviour 22, 193–222. Brown, S.D., Dooling, R.J., 1992. Perception of conspecific faces by budgerigars (Melopsittacus undulatus): I. Natural faces. J. Comp. Psychol. 106, 203–216. Clark, D.L., Uetz, G.W., 1992. Morph-independent mate selection in a dimorphic jumping spider: demonstration of movement bias in female choice using video-controlled courtship behavior. Anim. Behav. 43, 247–254. D’Eath, R.B., 1998. Can video images imitate real stimuli in animal behaviour experiments? Biol. Rev. 73, 267–292. Dooling, R.J., 1986. Perception of vocal signals by budgerigars (Melopsittacus undulatus). Exp. Biol. 45, 195–218. Farabaugh, S.M., Linzenbold, A., Dooling, R.J., 1994. Vocal plasticity in Budgerigars (Melopsittacus undulatus): evidence for social factors in the learning of contact calls. J. Comp. Psychol. 108, 81–92. Fleishman, L.J., Endler, J.A., 2000. Some comments on visual perception and the use of video playback in animal behavior studies. Acta Ethologica 3, 15–27. Fujita, K., Matsuzawa, T., 1986. A new procedure to study the perceptual world of animals with sensory reinforcement: recognition of humans by a chimpanzee. Primates 27, 283–291. Gallup, A.C., Swartwood, L., Militello, J., Sackett, S., 2015. Experimental evidence of contagious yawning in budgerigars (Melopsittacus undulatus). Anim. Cognit. 18, 1051–1058. Gilbertson, D.W., 1975. Courtship as a reinforcement for key pecking in the pigeon, Columba livia. Anim. Behav. 23, 735–744. Golbeck, J., Neustaedter, C., 2012. Pet video chat: monitoring and interacting with dogs over distance. CHI EA ‘12 Ext. Abstr. Human Factors Comput. Syst., 211–220.

Hile, A.G., Striedter, G.F., 2000. Call convergence within groups of female budgerigars (Melopsittacus undulatus). Ethology 106, 1105–1114. Hile, A.G., Plummer, T.K., Striedter, G.F., 2000. Male vocal imitation produces call convergence during pair bonding in budgerigars, Melopsittacus undulatus. Anim. Behav. 59, 1209–1218. Ikebuchi, M., Okanoya, K., 1999. Male zebra finches and Bengalese finches emit directed songs to the video images of conspecific females projected onto a TFT display. Zool. Sci. 16, 63–70. King, S.L., 2015. You talkin’to me? Interactive playback is a powerful yet underused tool in animal communication research. Biol. Lett. 11 (20150403). Kodric-Brown, A., Nicoletto, P.F., 1997. Repeatability of female choice in the guppy: response to live and videotaped males. Anim. Behav. 54, 369–376. Linton, S.J., 1981. A failure to establish key pecking for social reinforcement. J. Gen. Psychol. 104, 307–308. Mottley, K., Heyes, C., 2003. Budgerigars (Melopsittacus undulatus) copy virtual demonstrators in a two-action test. J. Comp. Psychol. 117, 363–370. Mui, R., Haselgrove, M., Pearce, J., Heyes, C., 2008. Automatic imitation in budgerigars. Proc. R. Soc. Lond. B Biol. 275, 2547–2553. Murata, K., Usui, K., Shibuya, Y., 2014. Effect of haptic perception on remote human-pet interaction. In: Yamamoto, S. (Ed.), Human Interface and the Management of Information. Information and Knowledge Design and Evaluation. Springer International Publishing, Berlin, pp. 226–232. Neustaedter, C., Golbeck, J., 2013. Exploring pet video chat: the remote awareness and interaction needs of families with dogs and cats. CSCW ‘13Proceedings of 2013 Conference on Computer Supported Cooperative Work, 1549–1554. Ophir, A.G., Galef, B.G., 2003. Female Japanese quail affiliate with live males that they have seen mate on video. Anim. Behav. 66, 369–375. Ord, T.J., Evans, C.S., 2002. Interactive video playback and opponent assessment in lizards. Behav. Process. 59, 55–65. Palagi, E., Leone, A., Mancini, G., Ferrari, P.F., 2009. Contagious yawning in gelada baboons as a possible expression of empathy. Proc. Natl. Acad. Sci. U. S. A. 106, 19262–19267. Parr, L.A., Winslow, J.T., Hopkins, W.D., de Waal, F., 2000. Recognizing facial cues: individual discrimination by chimpanzees (Pan troglodytes) and rhesus monkeys (Macaca mulatta). J. Comp. Psychol. 114, 47–60. Patton, T.B., Szafranski, G., Shimizu, T., 2010. Male pigeons react differentially to altered facial features of female pigeons. Behaviour 147, 757–773. Paukner, A., Anderson, J.R., 2006. Video-induced yawning in stumptail macaques (Macaca arctoides). Biol. Lett. 2, 36–38. Paxton, A., Dale, R., 2013. Frame-differencing methods for measuring bodily synchrony in conversation. Behav. Res. Methods 45, 329–343. Perret, A., Henry, L., Coulon, M., Caudal, J.P., Richard, J.P., Cousillas, H., Hausberger, M., George, I., 2015. Social visual contact, a primary drive for social animals? Anim. Cognit. 18, 657–666. Racca, A., Amadei, E., Ligout, S., Guo, K., Meints, K., Mills, D., 2010. Discrimination of human and dog faces and inversion responses in domestic dogs (Canis familiaris). Anim. Cognit. 13, 525–533. Rieucau, G., Giraldeau, L.A., 2009. Video playback and social foraging: simulated companions produce the group size effect in nutmeg mannikins. Anim. Behav. 78, 961–966. Rowland, W.J., Bolyard, K.J., Jenkins, J.J., Fowler, J., 1995. Video playback experiments on stickleback mate choice: female motivation and attentiveness to male colour cues. Anim. Behav. 49, 1559–1567. Shimizu, T., 1998. Conspecific recognition in pigeons (Columba livia) using dynamic video images. Behaviour 135, 43–53. Swartz, K.B., Rosenblum, L.A., 1980. Operant responding by bonnet macaques for color videotape recordings of social stimuli. Anim. Learn. Behav. 8, 311–321. Takahashi, M., Okanoya, K., 2013. An invisible sign stimulus: completion of occluded visual images in the Bengalese finch in an ecological context. Neuroreport 24, 370–374. Takahashi, M., Ikebuchi, M., Okanoya, K., 2005. Spatiotemporal properties of visual stimuli for song induction in Bengalese finches. Neuroreport 16, 1339–1343. Toda, K., Watanabe, S., 2008. Discrimination of moving video images of self by pigeons (Columba livia). Anim. Cognit. 11, 699–705. Trillmich, F., 1976. Spatial proximity and mate-specific behavior in a flock of budgerigars (Melopsittacus undulatus; Aves, Psittacidae). Z. Tierpsychol. 41, 307–331. Tschacher, W., Rees, G.M., Ramseyer, F., 2014. Nonverbal synchrony and affect in dyadic interactions. Front. Psychol. 5. Tsuchida, J., Izumi, A., 2009. The effects of age and sex on interest toward movies of conspecifics in Japanese macaques (Macaca fuscata). J. Am. Assoc. Lab. Anim. Sci. 48, 286–291. Uetz, G.W., Smith, E.I., 1999. Asymmetry in a visual signaling character and sexual selection in a wolf spider. Behav. Ecol. Sociobiol. 45, 87–93. Watanabe, S., Troje, N.F., 2006. Towards a virtual pigeon: a new technique for investigating avian social perception. Anim. Cognit. 9, 271–279. Wyndham, E., 1980. Diurnal cycle, behavior and social organization of the budgerigar Melopsittacus undulatus. Emu 80, 25–33.