This item is only available as the following downloads:
USING YOUR MELON: THE EFFECTS OF OBJECT RECOGNITION ON THE ECHOLOCATION OF AN ATLANTIC BOTTLENOSE DOLPHIN ( TURSIOPS TRUNCATUS ) BY KATHERINE NEWTON A Thesis Submitted to the Division of Social Sciences New College of Florida in partial fulfillment of the requirements for the degree Bachelor of Arts Under the sponsorship of Heidi Harley Sarasota, Florida May 2013
i Dedication To Mom and Dad, for instilling within me a love of animals and for encouraging me to follow my dreams and find happiness.
ii Acknowledgements This thesis could not have been done without a lot of support and encouragement. I would like to say thank you to: Professor Harley for always going above and beyond I do not believe words can accurately describe everything you have help ed me with and done for me in my years at New College Meeting with you was always reassuring, whether or not we figured out a n immediate solution to my problems You are a hu ge source of inspiration to me. Professors Bauer and Barton for not only being insightful and critical committee members, but also great professors who have expanded my knowledge and have taught me to think critically, both of which I hope to never take for granted. Wendi, for giving me access to the recordings and data, and always providing sound advice and ways to troubleshoot any issues. And your graph making skills are amazing. Diana for being much more savvy with technology than I am. Professor Cooper, for your eager ass istance with statistics. Calvin, for being an exceptional subject and a truly amazing animal. *** My family who are always there with words of wisdom encouragement, and positivity. My friends, for always having my back. You know who you are, b ut especially: Stephanie for being by my side through everything dolphin related that we encountered at New Co llege, Jennie and Judy, for helping me to find my academic niche at Ne w College Nicole, Lauren, and Alex, for frequently joining me in bitter thesis student ti rades, Roger for your clever way with words, And Norm for everything. Thank you for your endless patience, your constant motivation and your insane ability to keep me in high spirits. You have listened to everything I have had to say f rom my emotional rants, nonsensical logic, and frustrations to my excitements, happy thoughts, and bouts of childish glee. I sincerely appreciate all that you do for me, and I love having you by my side.
iii "For instance, on the planet Earth, man had al ways assumed that he was more intelligent than dolphins because he had achieved so much the wheel, New York, wars and so on whilst all the dolphins had ever done was muck about in the water having a good time. But conversely, the dolphins had always be lieved that they were far more intelligent than man for precisely the same reasons." Douglas Adams, The Hitchhiker's Guide to the Galaxy
iv TABLE OF CONTENTS Dedication ... i Acknowledgements ii Table of Contents .. iv Table of Tables and Figure s v Abstract vii Introduction 1 Method 12 Results .. 16 Discussion 19 References 26
v TABLE OF TABLES AND FIGURES Table 1 ..... 33 Table 2 34 Figure 1 : Dolphin Vocalizations .. 35 Figure 2 : The Facility ... 36 Figure 3 : The Object Sets 37 Figure 4 : The Task Set Up ... 38 Figure 5 : Click Count ing 39 Figure 6 : Average Number of Clicks With Respect to Famili arity and Accuracy .. 40 Figure 7 : Time Spent Echolocating With Respect to Familiarity and Accura cy 41 Figure 8 : Number of Click Trains With Respect to Fa miliarity and Accuracy ... 42
vi Figure 9 : Occurrence of the Terminal Burst Pulse With Respect to Familiarity and Accuracy .. 43
vii USING YOUR MELON: THE EFFECTS OF OBJECT RECOGNITION ON THE ECHOLOCATION OF A BOTTLENOSE DOLPHIN ( TURSIOPS TRUNCATUS ) Katherine Newton New College of Florida, 2013 ABSTRACT Dolphins use echolocation to discriminate amongst stimuli within their environment. Echolocation provides a unique way to learn about object recognition strategies because it is active (i.e., the dolphin produces clicks to gain information about objects) A few studies show that dolphins adapt their echolocation across contexts, but the characteristics that moti vate these adaptations are not clear The present study examines how an Atlantic bottlenose dolphin's ( Tursiops truncatus ) echoic investigation of stimuli changes across an object recognition task as initially novel objects become familiar and as performan ce accuracy increases. Acoustical and video rec ordings were obtained during a three alternative matc hing to sample task in which a blind folded dolphin examined a sample object and select ed the matching object from three alternatives. Analysis of the inves t igation of the sample object determined the number of clicks emitted, time spent echolocating the sample, number of echolocation trains, and occurrence of a terminal burst pulse. Results showed that number of clicks and time spent echolocating decreased w hen familiar objects were more easily recognized (i.e.,
viii performance accuracy was high ), suggesting that dolphins become faster and more efficient at echoic processing when objects are easily identified. Though there was low variability in the number of ech olocation trains, the time between echolocation trains may be important to processing previous echoes. The frequent occurrence of the terminal burst pulse suggests that this vocalization may function in some way during object recognition. Results also show ed that echoic effort stayed high when objects were difficult to identify, suggesting that the dolphin was putting in extra effort in order to try and solve the task. These data could inform training by the US Navy of bottlenose dolphins that recover and l ocate objects: because echoic efficiency increases with familiar, easily discriminated objects, the training program should include experience with targets of interest. ______________________________ Heidi E. Harle y Division of Social Sciences
Running head: EFFECTS OF OBJECT RECOGNITION ON ECHOLOCATION 1 Using Your Melon: The Effects of Object Re cognition on the Echolocation of an Atlantic Bottlenose Dolphin ( Tursiops truncatus ) Studying the ways in which animal s use their sensory systems provides valuable insight into the perceptions, motivations, and thought processes of many species. Bottlenos e dolphins ( Tursiops truncatus ) are an interesting species of study due to the diverse, complex, and foreign na ture of their vocal repertoire Dolphins rely on their acoustical processing for monitoring their environment and interacting with conspecifics. Their vocalizations are emitted at hi gh frequencies their hearing is sensitive to high frequency sounds and their processing of received acoustical information occurs very quickly. Dolphins' impressive use of acoustical information has been recognized by the United States Navy, which has an established Marine Mammal Program that is dedicated to studying, training, and deploying bottlenose dolphins and California sea lions ( Zalophus californianus ) to complete tasks such as marking and retrieving objects in the ocean (http://www.public.navy.mil/spawar/Pacific/71500/Pages/default.aspx) Because dolphins possess biological sonar (known as echolocation), they are essential to these tasks, as they can dive to great depths and use their echolocation to locate o bjects in acoustically and visually complex conditions. The Navy Marine Mammal Program also trains dolphins to detect and mark mine like objects and to help in guiding ships through safe passages of shallow water to help with landing troops ashore However, t hough there is a good amount of information known about dolphin echolocation, there are many
EFFECTS OF OBJECT RECOGNITION ON ECHOLOCATION 2 questions that are still unanswered. For example, we do not know much about the factors that affect changes in dolphin's echoic investigation of objects the topic of this thesis. Dolphin Vocalizations The vocal repertoire of dolphins is not limited to echolocation alone. Due to constraints in visibility, olfaction, and other senses affected by the marine environment, the acoustic channel has become the m ost important sense for dolphins in terms of social behaviors, foraging, and investigatio n of their environment. Bottlenose d olphins (hereafter referred to as "dolphins") have evolved to have a complex acoustical system, with a hearing range from about 50 Hz to more than 150 kHz (Johnson, 1967), matching the production capacity of their vocalizations and allowing for the perception of a broad range of frequencies. Dolphin vocalizations are often categorized in three ways: whistles, burst pulses, and echoloc ation. See Figure 1 for spectrograms of whistles, burst pulses, and echolocation. Whistles Whistles are a narrowband (i.e., the energy of the sound is limited to a narrow frequency range) vocalization with varying numbers of harmonics (i.e., sounds which occur at integer multiples of the fundamental frequency) and frequency modulations (i.e., ch anges in frequency within whistles). The fundamental frequency of whistles ranges from 0.8 to 24kHz, with a majority of frequencies occurring within 3.5 14.5kHz (Richardson, Greene, Malme, & Thomson, 1995 ). Each dolphin possesses an individually distincti ve whistle, termed a "signature whistle" that has a specific frequency modulation pattern and can remain consistent for a lifetime (Caldwell & Caldwell, 1965, 1968; Sayigh, Tyack, Wells, & Scott, 1990).
EFFECTS OF OBJECT RECOGNITION ON ECHOLOCATION 3 Dolphins often use their signature whistles when they are separated from individuals in their group, suggesting that these whistles serve as a contact call to maintain group cohesion (Janik & Slater, 1998; Watwood, Owen, Tyack, & Wells, 2005). Additionally, signature whistles have been observed in mother/cal f reunions, alloparental (i.e., care for a calf by an individual other than the parent) care, and courtship (Herzing, 1996). When developing their signature whistles, female calves tend to produce a signature whistle that is distinctive from that of their mothers, while male calves produce one more similar to that of their mothers (Sayigh, Tyack, Wells, & Scott, 1990). Infant whistle repertoires are quite similar to that of their social group, suggesting that dolphins possess vocal plasticity and learning (McCowan & Reiss, 1995). Additionally, as males age, their whistle repertoires expand (Sayigh, Tyack, Wells, & Scott, 1990), and they begin sharing their signature whistle with their male partner s (Watwood, Tyack, & Wells, 2004). Dolphins are also capable of matching whistles such that, on occasion, an individual will respond to the whistle of a conspecific by emitting the same type of whistle, suggesting that whistles may be used to address one another (Janik, 2000). Burst Pulses Like whistles, burst puls es are social sounds, but they are broadband sounds in which many frequencies are produced at once. These sounds are often described as buzzes, squawks, screams, barks, and yelps (Herzing, 1996; Wood, 1953). Burst pulses often possess a large amount of ene rgy between 60 and 150 kHz, although energy below 20 kHz has been observed to increase as aggressive behavior increases (Bloomqvist & Amundin, 2004). Burst pulses are very common in the context of aggressive behavior,
EFFECTS OF OBJECT RECOGNITION ON ECHOLOCATION 4 but also occur during excitement, dist ress, courtship, discipline, sexual play, and foraging (Bloomqvist & Amundin, 2004; Herzing, 1996; Overstrom, 1983). Echolocation Another broadband vocalization is echolocation clicks. These are used for biosonar to obtain information about the identity location, and characteristics of objects, even in cluttered and noisy environments (see Au, 1993, for a review). Clicks are short in duration (between 40 and 70 !s) with peak frequencies between 30 and 135 kHz. Clicks originate in the dolphin's forehead by passing air through a nasal structure known as the "monkey lips" (Cranford, 2000). The emitted clicks are reflected off surfaces in the environment, and these echoes are received in the dolphin's jaw where they are then conducted to the inner ear (Brill Sevenich, Sullivan, Sustman, & Witt, 1988). Dolphins wait to receive a returning echo before emitting another click ; variance in the interv al between clicks is predicted by distance between the dolphin an d the object it is echolocating (Au, Floyd, Penner & Murchison, 1974). Dolphins have an auditory integration time (i.e., the amount of time it takes for the dolphin to tell two successive clicks apart) of about 264 !s (Au, Moore, & Pawloski, 1988; Moore, Hall, Friedl, & Nachtigall, 1984). Behavioral st udies have found that echolocation is used in foraging efforts (Herzing, 1996) as well as for inspecting novel objects (Wood, 1953). Xitco and Roitblat (1996) have gathered data that suggest that dolphins can gather information about objects by listening t o the returning echoes of a n echolocating companion, indicating that echolocation information may be shared between neighbors. This evidence suggests that the production of echolocation clicks is not necessarily an integral part in the processing
EFFECTS OF OBJECT RECOGNITION ON ECHOLOCATION 5 of echo features since a non echolocating dolphin can process a neighbor's echoes (Xitco & Roitblat, 1996). Many studies on dolphin echolocation have sought to determine echolocation capabilities in varying contexts. These studies often involve target detection tasks, in which the dolphin indicates whet her a specified target is either present or absent. In such tasks, the dolphin is both emitting clicks and receiving echoes and uses this gathered information to make its selection. Target detection studies have found that dolphins can detect targets in th e presence of background noise, and that dolphins' accuracy at target detection decreases as background noise levels exceed 77dB re 1!Pa (Au & Penner, 1981; Au, Penner, & Kadane, 1982). Not only does accuracy decrease at this noise level, but the number of clicks emitted also decrease s suggesting that dolphins reduce their efforts to detect a target when the task becomes too difficult to solve (Au, Penner, & Kadane, 1982). Dolphins are also able to echolocate targets at varying distances. Au and Snyder (19 80) found that a dolphin was successful at detecting a target object up to 95 meters away. Performance declined as the distance increased past 95 meters, with 113 meters being the distance where accuracy was at chance level (i.e., 50%). Additionally, dolph ins can discriminate between objects composed of different materials. A dolphin was able to successfully discriminate between objects made from aluminum (the standard target), versus steel and coral r ock, even when these objects were presented at varying a ngles relative to the dolphin (Au & Turl, 1991). Acoustic detection tasks are not limited to dolphins echolocating a target object. Auditory stimuli have also been used in target detection, i.e., a task in which the dolph in listens to and identifies an au ditory st imulus that contains some specific acoustical
EFFECTS OF OBJECT RECOGNITION ON ECHOLOCATION 6 feature. In one such task, a dolphin listened to synthetic echolocation trains and had to indicate whether amplitude modulation was or was not present in each train. The dolphin was able to discr iminat e between the echo trains up to an amplitude variation of 0.8 dB re 1 !Pa between minimum and maximum suggesting that they are se nsitive to changes in amplitude (Dankiewicz, Helweg, Moore, & Zafran, 2002). Object Representation The echoes that dolphins re ceive possess a variety of acoustic features that relay information about the echolocated object. The acoustic features of echoes (e.g., target strength, highlights, peak frequency, and center frequency) that dolp hins use to recognize objects are not entir ely known, though they likely use multiple features as opposed to one single feature. Using multiple characteristics is advantageous because values of some characteristics could disambiguate others, and characteristics change across objects within the envi ronment (DeLong, Au, Lemonds, Harley, & Roitblat, 2006). Although dolphins have greater sensitivity for high frequencies than humans (Thompson & Herman, 1975), their inner ear functions similarly to that of humans, suggesting that insight into dolphins' u se of echo features may be achieved by using human listeners. For example, in one study h uman participants listened to echoes that had been slowed down so that humans could hear them. The echoes were recorded from objects that a dolphin had correctly match ed within a matching task and were recor ded at varying aspect angles, because the dolphin echolocated the objects at different angles. The human listeners discriminated between the echoes and answer ed questions as to how they were discriminating The participants reported that changes in loudness, pitch, and timbre across echolocation trains were important for discriminating objects from varying
EFFECTS OF OBJECT RECOGNITION ON ECHOLOCATION 7 orientation s For some of the objects (particularly those from different orientations), the dolphin's error patterns matched that of the humans', implying that the dolphin may have used the same cues reported by the humans (DeLong, Au, Harley, Roitblat, & Pytka, 2007). Cross modal match to sample (MTS) studies provide evidence suggesting that dolphin s are able to represent objects both visually and echoically and transfer given information about an object from one modality to the other. Cross modal MTS is achieved in dolphin research by testing the modalities of both echolocation and vision such that perce ption of an object in one modality results in the recognition and matching of that object in the other modality. Dolphins have generally been successful in cross modal matching, suggesting that they can learn about visual and echoic object features and can use either modality to recognize an object (Harley, Roitblat, & Nachtigall, 1996). Such cross modal su ccess has been found with unfamiliar objects : Pack and Herman (1995) found high performance accuracy in an MTS task (both within and across modalities) d uring the first matching trial with each object suggesting that the matching was not the result of learned associations but rather immediate recognition (Herman, Pack, & Hoffman Kuhnt, 1998; 1995). Harley, Putman, and Roitblat (2003) conducted a cross mod al experiment in which the dolphin was rewarded for matching a sample object to a non identical alternative (i.e., after sensing object A, choosing object B is rewarded and vice versa). Although the dolphin was trained and rewarded in choosing the non iden tical alternative it chose the identical alternative 74.1% of the time. These results suggest that dolphins do
EFFECTS OF OBJECT RECOGNITION ON ECHOLOCATION 8 not learn about and later recognize objects due to a history of being rewarded, but rather they immediately recognize objects due to obtained se nsory information In addition to transferring information from one sense to another, dolphins can also integrate information across echolocation trains, as shown in studies involving change across object orientation. Echoes change as object orientation ch anges, and so Helweg, Au, Roitblat, and Nachtigall (1996) measured a dolphin's ability to match objects at different angles (i.e., the objects were free to rotate) an d recorded the dolphin's clicks and echoes Their recordings showed that the echoes that t he dolphin received from the target s changed as the objects rotated, yet the dolphin continued to successfully match the objects. This suggests that dolphins are able to integrate changes in returning echo features across consecutive echoes. Dolphins also have a well developed short term memory for auditory stimuli. Studies in which a dolphin matches auditory stimuli across a short delay between the presentation of the sample and alternatives have shown that dolphins are successful at matching the stimuli for delays of between two and four minutes (Herman & Gordon, 1974; Herman & Thompson, 1982). In conclusion, the dolph in's ability to transfer information about objects between vision and echolocation, to integrate information across multiple echoes and echolocation trains, and to retain auditory stimuli within short term memory suggests that dolphins convert sound into representations of objects (see Roitblat, 2002, for a review). Echoic Analysis Evidence not only suggests that dolphins utilize and integrate multiple features and changes in features of returning echoes, but they also possess the ability to adapt and
EFFECTS OF OBJECT RECOGNITION ON ECHOLOCATION 9 control the structure of the clicks that they emit (Houser, Helweg, & Moore, 1999). F or example, Au, Floyd, Penner, and Murchison (1974) measured the echolocation clicks of dolphins at fixed distances from a target using a within subjects design The amount of time between each click (the inter click interval) varied over trials of the same distance suggesting that the variability in inter click interval s i s not only caused by changes in target distance. In addition, there were small fluctu ations in the amplitude of the clicks. Houser, Helweg, and Moore (1999) also noted differences in emitted clicks (e.g., frequency range, frequency distri bution, maximum frequencies, and amplitude) between dolphins par ticipating in the same tasks while Houser et al. (2005) found that the degree of success of finding objects varied as a function of distinct search strategies: one dolphin echolocated frequent ly and only responded that an object was present if it was actually detected, but the second dolphin minimized energy resources by only echolocating when necessary, i.e., when it did not have enough information to make a response. Analysis of echoic invest igation in matching tasks provides insight into how dolphins echolocate objects that they are able t o recognize. Roitblat, Penner, and Nachtigall (1990) examined the matching performance of an echolocating dolphin wearing soft, latex eyecups to ensure tha t the objects to be matched could not be seen (i.e., the dolphin could only sense them via echolocation). During each trial, the dolphin echolocated the sample followed by three alternatives, and received fish for selecting the alternative that matched the sample. Accuracy averaged 94.5% correct over 48 sessions with three objects. The researchers investigated the dolphin's expended effort in examining each stimulus. The dolphin emitted an average of 37.2 clicks while investigating the sample. When echoloca ting the alternatives, the dolphin always scanned
EFFECTS OF OBJECT RECOGNITION ON ECHOLOCATION 10 the left object first, then the middle, then the right. The number of clicks emitted depended on the location and identity of the matching stimulus, leading the researchers to suggest that the dolphin put i n more effort in order to obtain higher accuracy because the dolphin emitted more clicks for the objects that were more difficult to identify. The researchers postulated a sequential sampling computational model of the echolocation and decision processes o f the dolphin, in which each click gives rise to an echo from one of three overlapping distributions, one corresponding to the matching comparison and one for each of the two mismatching comparisons. These distributions overlap such that the matching compa rison is in the center with the mismatches on either side. One vertical line through these distributions correlates to one echo, and additional echoes are distributed according to one of the three distributions. The average of these echoes computes a likel ihood ratio, which is the ratio of the probability that the echo came from the matching distribution to the probability that it came from a non matching distribution. Based on this model, the dolphin uses the ratio to make a Bayesian decision regarding the identity of the scanned stimulus (Roitblat, Penner, & Nachtigall, 1990). The study by Roitblat, Penner, and Nachtigall (1990) was compelling in terms of the analysis of the dolp hin's active use of echolocation to investigate objects. However, they prese nted results from the final 48 sessions the dolphin had with the objects, meaning that at the time of echoic analysis the dolphin was matching the objects at a high accuracy Xitco and Roitblat (1996) noted that matching performance beco mes better with suc ce ssive trials with stimuli. B ecause dolphins vary their echoic investigation strategies during target detection (Houser et al., 2005), data concerning how echoic investigation may change with respect to object familiarity and matching accuracy may
EFFECTS OF OBJECT RECOGNITION ON ECHOLOCATION 11 provide further insight into how dolphins represent objects through echolocation. The goal of the current study was to gather that data. The Current Study Dolphins have a complex vocalization system made up of whistles, burst pulses, and echolocation clicks (Jani k, 2009). Their echolocation system is particularly sophisticated. They process sounds quickly (Au, Moore, & Pawloski, 1988; Moor e Hall, Friedl, & Nachtigall, 1984) and remember them well (Herman & Gordon, 1974; Herman & Thompson, 1982). Dolphins can echo locate and detect objects up to 95 meters away (Au & Snyder, 1980) and can detect objects within the presence of noise, although they become less accurate and reduce their efforts as noise increases past 77dB (Au & Penner, 1981; Au, Penner, & Kadane, 1982) Dolphins can change and adapt click characteristics, such as frequency, amplitude, and interval, while echolocating within different contexts (Au, Floyd, Penner, & Murchison, 1974; Harley & DeLong, 2008; Houser, Helweg, & Moore, 1999), and can also inte grate auditory and visual information about objects (Harley, Roitblat, & Nachtigall, 1996; Herman, Pack, & Hoffmann Kuhnt, 1998; Pack & Herman, 1995). Echoic information gained from various orientations and aspect angles of objects can also be integrated ( Au & Turl, 1991; Helweg, Au, Roitblat, & Nachtigall, 1996). Houser et al. (2005) found that dolphins have varying search strategies in target detection, and Roitblat, Penner, and Nachtigall (1990) found that, although dolphins can become very good at matching objects, they often emitted more clicks when objects were difficult to identify. Despite the large amount of research on object recognition and
EFFECTS OF OBJECT RECOGNITION ON ECHOLOCATION 12 matching tasks, echoic investigation of objects during an MTS task as the dolphin gains experience with unfamiliar objects has not been analyzed. The current study aims to examine how a dolphin's echoic investigation of objec ts in a MTS task changes over levels of object familiarity and performance accurac y across several object sets Analysis will be conducted using acoustic and video files of a dolphin performing an echoic MTS task and will examine c hanges in the number of clicks and the amount of time echolocating the sample objects across accuracy fami liarity, and object type If, as suggested by previous research, the dolph in is able recognize objects more accurately over time and increase searching efficiency, then the number of clicks and the amount of time echolocating may decrease as objects become more familiar and more easily recognized. Such results could imply that object recognition yields faster and more efficient echo ic processing. Method Subject The subject was a male Atlantic bottlenose dolphin ( Tursiops truncatus ) born in 1994 at a facil ity in the Florida Keys and moved to his current facility in 2003. The subject's hea ring was determined (via audito ry evoked potential) to be normal for his species (W. Fellner, personal communication, 4/8/13). He and his three tank mates had participated in previous cognitive research (see Harley, Fellner, & Stamper, 2010).
EFFECTS OF OBJECT RECOGNITION ON ECHOLOCATION 13 Facility The subject and his tank mates resided in a large tank consisting of 22 million liters of salt water made up of three pools (Main Tank, A Pool, and B Pool) connected by gat es and a canal. The Main Tank is the largest of the three and holds a variety of aquatic life that is on exhibit for the public. The current study was conducted within B Pool, which measured 8.2m long by 7m wide by 2.1m deep. See Figure 2 for a diagram of the facility. Materials Many objects served as stimuli in the study. See Figure 3 for a display of each stimulus set. Obje ct sets are named based on the dolphin's increase in accuracy over five sessions with each object set as well as the dolphin's overall performance accuracy The object set names thus refer to: an 11% rise to 44% performance accuracy (hereafter 11% Poor), a 28% rise to 50% performance accuracy (hereafter 28% Poor), an 11% rise to 100% performance accuracy (hereafter 11% Good), a nd a 28% rise to 78% performance accuracy (hereafter 28% Good). The subject was initially unfamiliar with each stimulus but became familiar with objects across five 18 trial sessions. Although several hydrophones recorded the dolphin's vocalizations, only the clicks recorded by the hydrophone positioned near the sample were analyzed. The hydrophones used to record the vocalizations were High Tech, Inc. HTI 96MIN hydrophones with a flat frequency response of 2Hz to 30kHz (though the actual recording range was 0Hz to 50kHz), and clicks were recorded at a sampling rate of 100,000Hz per second. The clicks were recorded onto a Lenovo T410 laptop computer using Avisoft RECORDER USG version 4.2.8 (http://www.avisoft.com).
EFFECTS OF OBJECT RECOGNITION ON ECHOLOCATION 14 Video was recorded using a PC Osprey 4 channel video card with H.264 Webcam software that simultaneously recorded the surface of A and B Pools, above the water in the Main Tank, and below the water in the Main Tank. For this study, only the video recordings of the surface of A and B Pool were u sed. The audio recordings were analyzed using Avisoft SASlab Pro Sound Analysis and Synthesis Laboratory version 5.2.01. All recordings (as .wav files) were converted into spectrograms, and audio playback was conducted both in real time and slowed by 48%. Procedure At the start of the study, the subject already performed capably in a three alternative match to sample (MTS) task. He wore soft, latex eyecups during trials to preclude visual cues. He could pop the eyecups off at will, but he was trained to wear them. If an eyecup came off during a trial, stimuli were immediately pulled from the tank. If he had been near any stimuli they were no longer used in the study. A trial began when the dolphin positioned himself in front of his trainer, who the n signaled him tact ilely to swim 6.1m to the sample object. The dolphin could swim at his own p ace ( X = 7.36s, SD = 1.6), though once he reached the sample a research assistant pulled it out of the water after a certain amount of t ime ( X = 2.28s, SD = .78). Although the research assistant determined how long the sample would remain in the water after the dolphin reached it which ranged from 1.61s to 2.96s the dolphin could control his access to the sample by changing his swim times, which had a large range of 4s to 13s Once the sample was pulled out of the water, the dolphin swam 6.1m to an array of three alternatives in which there was an object identical to the sample. The dolphin indicated
EFFECTS OF OBJECT RECOGNITION ON ECHOLOCATION 15 his choice by positioning himself in front of an object with his rostrum pointing towards it and vocalizing. A research assistant nave to the id entity of the sample identified his choice. If the choice object matched the sample, the trainer blew a whistle after which the dolphin received one to two capelin. If the dolphin chose an object that did not match the sample, the trainer slapped the wall and the dolphin positioned himself in front of his trainer for the next trial. Each session contained 18 trials Across those 18 trials, each of the three objects was the sample the same number of times. The location of the alternatives was also balanced such that each object was presented in each location the same number of trials, both when it was the correct choice and when it was an incorrect choice All objects were suspended 0.7m from the walls of the tank, with 1.2m between each alternative. Objects were suspended such that the center of each object was 40.6cm under the surface of the water. See Figure 4 for a representation of the task set up. Analysis Sessions were analyzed based on the stimuli's familiarity and the accuracy with which the dolphin matched. Therefore, first sessions with unfamiliar objects and fifth sessions with familiar objects were analyzed. In addition, sessions on which the dolphin produced his lowest and highest performance accuracy were also analyzed. Table 1 presents performa nce accuracy across all sessions; analyzed sessions are identified. Echoic examinations of the sample objects recordings were assessed via acoustic and video analysis. Video recordings and session notes taken by the on site researchers were used to synch ronize behaviors with the audio recordings. Video recordings were also used to determine the amount of time it took the dolphin to swim towards the sample
EFFECTS OF OBJECT RECOGNITION ON ECHOLOCATION 16 as well as the length of time the sample was in the water. This time varied based on the dolphin's sw imming speed and how quickly he reached the sample. Reaching the sample was defined as the time the dolphin's rostrum was directly in front of and no more than 0.3m away from the sample. Click trains were measured for length of train in time and number of clicks. Length of time was defined as the amount of time elapsed from the first click to the last click in a given train, and the number of clicks was defined as the number of clicks in each click train. Clicks were counted using the Pulse Train Analysis f eature on Avisoft. However, this feature did not work with every click train (e.g., background noise sometimes masked the clicks), and therefore some click trains were counted by hand. See Figure 5 for an example of the Avisoft Pulse Train Analysis feature If the amount of time between each click was 0.01s or less, the vocal ization was categorized as a burst pulse (as in Lammers, Au, Aubauer, & Nachitgall, 2003) and, therefore, not included in these data. Burst pulses occurred most often at the end of a cl ick train, usually the final click train before the sample was pulled out of the water. If the amount of time between one click and a nother was 0.3s or greater, trains were categorized as being two different click trains. Results Echoic effort differenc es between object sets that were difficult to discriminate (11% Poor and 28% Poor, hereafter poor object sets) and object sets that were easy to discriminate (11% Good and 28% Good, hereafter good object sets) were analyzed in relation to object familiarit y, with session 1 being unfamiliar and session 5 being familiar. Echoic effort was measured by: the number of emitted clicks directed to the
EFFECTS OF OBJECT RECOGNITION ON ECHOLOCATION 17 sample object, time spent echolocating the sample, number of click trains completed, and the occurrence of the term inal burst pulse. See Table 2 for a complete summary of the sessions analyzed and the results. There were a few instances in which certain trials within object sets could not be utilized. Trials 10 18 of session 1 of Object Set 11% Poor and trials 1 and 2 of session 5 for Object Set 28% Poor were not recorded via hydrophone, and therefore were not analyzed. The beginning of trial 1 of session 2 of Object Set 11% Good was not fully recorded, and therefore was not analyzed. Number of Emitted Clicks Clicks were counted as soon as the dolphin left his trainer until the sample was pulled out of the water. For the poor object sets, the dolphin did not produce a significantly different number of clicks for the familiar objects in session 5 ( M = 236.94, SD = 76.1 1) versus the unfamiliar objects in session 1 ( M = 225.37, SD = 142.74), matched pair t (26 ) = .21, p = .84. For the good object sets, the dolphin produced significantly fewer clicks for the familiar objects in session 5 ( M = 162.81, SD = 68.01) versus the unfamiliar objects in session 1 ( M = 214.78, SD = 74.84), matched pair t (35) = 3.56, p = .001. As indicated by Table 2, there was a good deal of variability across sets. See Table 2 for number of clicks for specific object sets and Figure 6 for a graphic al representation of number of emitted clicks directed to the sample object for familiarity and accuracy. Time Spent Echolocating the Sample Time spent echolocating the sample was analyzed by measuring the tim e duration of each click train. For the poor ob ject sets, the dolphin did not spend a significantly
EFFECTS OF OBJECT RECOGNITION ON ECHOLOCATION 18 different amount of time echolocating the familiar objects in session 5 ( M = 5.13, SD = 1.57) versus the unfamiliar objects in session 1 ( M = 4.65, SD = 3.14), matched pair t (26 ) = .18, p = .86. For the good object sets, the dolphin spent significantly less time echolocating the familiar objects in session 5 ( M = 4.31, SD = 1.6) versus the unfamiliar objects in session 1 ( M = 5.13, SD = 1.84), matched pair t (35) = 2.16 p = .04 See Table 2 for time spe nt echolocating for specific object sets and Figure 7 for a graphical representation the time sp ent echolocating the sample for familiarity and accuracy Number of Click Trains During analysis, it was noted that the dolphin frequently emitted more than one click train, and the number of click trains varied from trial to trial. This variable was therefore analyzed. For the poor object sets, the dolphin emitted more echolocation trains for the familiar objects in session 5 ( M = 2.26, SD = 1.78) versus the unf amiliar objects in session 1 ( M = 1.78, SD = .93). For the good object sets, the dolphin emitted about the same number of echolocation trains for the familiar objects in session 5 ( M = 2.03, SD = .88) versus the unfamiliar objects in session 1 ( M = 2.17, S D = .88). Due to the low variability in number of echolocation trains, this variable may not be affected by object recognition. See Table 2 for number of click trains for specific object sets and Figure 8 for a graphical representation of the number of cli ck trains for familiarity and accuracy. Terminal Burst Pulses Another additional variable noted during analysis was the emission of a burst pulse sound at the end of inspecting the sample, typically called a terminal burst. Because the terminal burst was present more frequently than not ( 2 [1, N = 168] = 42, p = <.01), its use was analyzed. For the poor object sets, the dolphin emitted the burst pulse more
EFFECTS OF OBJECT RECOGNITION ON ECHOLOCATION 19 often for the familiar objects in session 5 ( M = 82.89%) versus the unfamiliar objects in session 1 ( M = 75%). For the good object sets, the dolphin emitted the burst pulse less often for the familiar objects in session 5 ( M = 63.89%) versus the unfamiliar objects in session 1 ( M = 77.78%). Due to the frequent occurrence of the terminal burst pulse, it m ay be that this vocalization may help in gathering further information about the objects, particularly objects that are more difficult to discriminate. See Table 2 for the occurrence of the terminal burst for specific object sets and Figure 9 for a graphic al representation of the occurrence of the terminal burst. Discussion In the present study, a blindfolded dolphin participated in a three alternative echoic match to sample (MTS) task w ith four different object sets. Prior evidence suggests that dolphin s can do this task well (Roitblat, Penner, & Nachtigall, 1990), and, indeed, the subject dolphin here was clearly good at echoic matching (as indicated by his high accuracy with Object Set 11% Good, M = 95.55%), although his accuracy depended on the diffic ulty of the object set. As the objects beca me more familiar between the first and fifth sessions, the dolphin emitted fewer clicks for object sets on which discrimination was good, but he emitted the same or more clicks for sets in which discrimination was poor Therefore familiarity did not guarantee a decrease in echoic investig ation because it only occurred with easily discriminable sets. For the object sets that were hardest to match, i.e., the sets with the lowe st performance accuracy, an increase in echolocation efforts occurred across sessions, thereby suggesting the difficulty of matching the objects had the strongest effect on how the dolphin
EFFECTS OF OBJECT RECOGNITION ON ECHOLOCATION 20 echolocated. These results are similar to those found by Roitblat, Penner, and Nachtigall (1990) who found that dolphins emit more clicks when objects are difficult to identify Familiarity does not have the same effect because it only caused less echolocation effort when accuracy also had an effect. The number of click trains did not vary much across object sets, suggesting that these differences may not be substantial enough to make a strong conclusion. However, the number of click trains slightly increased in several cases, suggesting that the dolphin may be utilizing the time when echolocation is not occu rring to process the information it is receiving. The terminal burst pulse occurred in a high percentage of trials across all object sets, even to the point where it occurred in every trial for session 1 of Object Set 11% Good. The high percentage of term inal burst pulses suggests that this vocalization must function in some way during object recognition. Implications These r esults suggest that dolphins change their echoic investigation of well recognized objects in favor of increase d efficiency. That is, the dolphin spends less time echolocating and emits fewer clicks when objects are familiar and easy to identify These results are consistent with the finding that dolphins often emit more clicks for objects that they have difficulty in identifying (Roitbl at, Penner, & Nachtigall, 1990). The suggestion that dolphins change the way in which they echolocate objects depending on the ease of object recognition is also consistent with the evidence that dolphins can change their echoic search strategy (H ouser et al., 2005).
EFFECTS OF OBJECT RECOGNITION ON ECHOLOCATION 21 The varying number of click trains is a new finding in dolphin echolocation research. These results suggest that the number of click trains the dolphin emits increases in most cases, meaning that the dolphin splits its echolocation into mu ltiple trains as opposed to consolidating it. Inter click intervals are often the length of two way travel time between the dolphin and the object (click to object, echo to dolphin) with the addition of a little extra time which may be used for processing of those echoes (Au, Floyd, Penner, & Murchison, 1974). Perhaps the time between the trains, when the dolphin is not echoloca ting, is also important to the dolphin's processing of previous echoes. The frequent occurrence of a terminal burst pulse at the end of the echoic investigation may co ntain important information on which the dolphin relies because it occurred in high percentages across all object sets Although there has been some behavioral research with burst pulses (Bloomqvist & Amundin, 2004; He rzing, 1996; Lilly & Miller, 1961; Overstrom, 1983), not much is known about the use of this vocalization in non social contexts such as object recognition. Based on the pr esent findings, it appears that the terminal burst may be an additional vocalization used for gaining information about the objects to be discriminated. Because the dolphin's performance accuracy for matching novel objects increases with exposure to the objects, it can be inferred that the task is initially a bottom up process, i.e., the dolphin begins with gathering information about the object and constructs this information into a perception. As the object becomes more familiar and the dolphin's performance accuracy increases, the task then includes more top down proces sing i.e., the dolphin begins with expectation s about the objects and uses them to
EFFECTS OF OBJECT RECOGNITION ON ECHOLOCATION 22 judge its perceptions. The dolphin is likely storing information about the objects within its memory and retrieving that information during echoic investigation using short term memory for information gathered in each trial (as found in previous auditory tasks with dolphins, Herman & Gordon, 1974; Herman & Thompson, 1982) and using long term memory for information gathered across sessions These findings are s imilar to those previously found with human participants. Jo licoeur (1985) and Shinar and Owen (1973) studied object recognition in humans After determining that the time it takes humans to recognize objects decreases with practice and exposure, they rota ted those familiar objects and displayed them to the human participants at varying angle s Once the familiar objects were distorted such that they were unfamiliar, the human participants took longer to recognize them and made more errors in their classific ation. Again, through practice with the distorted objects, recognition time and error rates declined. The human participants' knowledge and expectations about those objects change d caus ing processing to shift from bottom up to top down Similarly, the dol phin subject of the current study decreased echoic investigation of objects and had fewer errors as objects became more familiar. Suggestions for Future Research In the current study, the dolphin varied his swim times to the samples, and the trainer dicta ted removal times of the sample. Therefore, exposure times to the sample were uncontrolled, and we cannot determine the minimum exposure the dolphin needs in order to successfully complete the task. Future work in which exposure time is systematically vari ed would give us more information about the relation between echoic efficiency and matching accuracy.
EFFECTS OF OBJECT RECOGNITION ON ECHOLOCATION 23 In addition as noted by Houser et al. (2005), individual dolphins can vary in their echoic search strategies. The dolphin in this study decreased its echolocation efforts when performance accuracy was high, yet continued or increased its efforts when performance accurac y was either low or decreasing. Replication of the present study with other dolphins could provide insight into whether object recognition yields different or similar search strategies across individuals. The findings of the present study suggest that echoic investigation, in terms of time echolocating, emitted clicks, and number of click trains, changes with object recognition. Previous research shows that additional click characteristics, such as frequenc y and amplitude, can be manipulated by echolocating dolphins ( Houser, Helweg, & Moore, 1999). Therefore, future analysis of echoic investigation with respect to object recognitio n should analyze these features to determine how they may be affected across o bject sets. Roitblat, Penner, and Nachtigall (1990) found that their dolphin emitted an average of 37.2 clicks in a matching task, which is much lower than the average number of emitted clicks in this study (the lowest being 160.39 clicks). However, their study utilized highly recognizable objects with which the dolphin had had a large quantity of previous exposure Thus, it may be worthwhile to examine the changes in echoic investigation of an object set for many sess ions until matching accuracy is high ev ery time to determine if there is a threshold for the number of clicks the dolphin emits in order to recognize the object. Such a threshold could provide information about what the dolphin is using for recognition.
EFFECTS OF OBJECT RECOGNITION ON ECHOLOCATION 24 Also as suggested by this study, termina l click train burst pulses may result in valuable information in regards to object recognition T he use of burst pulses shoul d be further investigated to determine additional uses for this vocalization. Further research on echoic examination of objects wil l add to this and previous research, and thus would allow us to continue to learn and understand how dolphins represent objects and stimuli within their environment. Conclusion That dolphins reduce echoic effort when echolocating familiar and easily identified objects could prove beneficial to the United States Navy Marine Mammal Program, which uses dolphins to locate and mark mines and other objects in the water that are difficult for divers to find because of extreme depths and poor visual condition s (United States Navy, n.d.). If the dolphins are able to spend less time searching for objects due to their heightened recognition, then they can search more areas in a shorter amount of time, increasing their work output. Taking this into account, the do lphins should be trained in detecting and discriminating the objects for which they will be searching to the greatest extent possible. Addit ionally, the current study can help us to further understand how dolphins use echolocation for detecting stimuli wi thin their natural environment, such as prey. The suggestion that dolphins reduce echoic investigation for familiar, easily identified objects could imply that, with experience, dolphins become more efficient at echoically processing stimuli that they freq uently encounter Using these findings, as well as previous and future findings, we can hope to learn more about how dolphins use the
EFFECTS OF OBJECT RECOGNITION ON ECHOLOCATION 25 sensory system of echolocation in perceiving and interpreting their environment, and all of the stimuli within it.
EFFECTS OF OBJECT RECOGNITION ON ECHOLOCATION 26 References Au, W. W. L. (1993). The Sonar of Dolphins New York: Springer Verlag. Au, W.W.L., Floyd, R.W., Penner, R.H., & Murchison, A.E. (1974). Measurement of echolocation signals of the Atlantic bottlenose dolphin, Tursiops truncatus Montagu, in open waters. Journal of the Acoustical Society of America, 56 (4), 1280 1290. Au, W.W.L., Moore, P.W.B., & Pawloski, D.A. (1988). Detection of complex echoes in noise by an echolocating dolphin. Journal for the Acoustical Society of America, 83 (2), 662 668. Au, W.W.L. & Penner, R.H. (1981). Target detection in noise by echolocating Atlantic bottlenose dolphins. Journal of the Acoustical Society of America, 70 (3), 687 693. Au, W.W.L, Penner, R.H., & Kadane, J. (1982). Acoustic behavior of echolocating Atlantic Bottlenose Dolphins. Journal of the Acoustical Society of America, 71 (5), 1269 1275. Au, W.W.L. & Snyder, K.J. (1980). Long range target detection in open waters by an echolocating Atlantic Bottlenose dolphin ( Tursiops trunc atus ). Journal of the Acoustical Society of America, 68 (4), 1077 1084. Au, W.W.L. & Turl, C.W. (1991). Material composition discrimination of cylinders at different aspect angles by an echolocating dolphin. Journal of the Acoustical Society of America, 89 (5), 2448 2451.
EFFECTS OF OBJECT RECOGNITION ON ECHOLOCATION 27 Bloomqvist, C. & Amundin, M. (2004). High frequency burst pulse sounds in agonistic/aggressive interactions in bottlenose dolphins ( Tursiops truncatus ). In Thomas, J., Moss, C., & Vater, M. (Eds.), Echolocation i n Bats and Dolphins (pp. 425 431). Chicago: University of Chicago Press. Brill, R. L., Sevenich, M. L., Sullivan, T. J. Sustman, J. D., & Witt, R. E. (1988). Behavioral evidence for hearing through the lower jaw by an echolocating dolphin ( Tursiops trun catus ). Marine Mammal Science, 4 223 230. Caldwell, M. C. & Caldwell, D. K. (1965). Invidualized whistle contours in bottlenose dolphins ( Tursiops truncatus ). Nature, 207 434 435. Caldwell, M. C. & Caldwell, D. K. (1968). Vocalization of nave dolphins in small groups. Science, 159 (3819), 1121 1123. Cranford, T. W. (2000). In search of impulse sound sources in odontocetes. In W. W. L. Au, A. N. Popper, & R. R. Fay (Eds.), Hearing by Whales and Dolphins (pp. 109 155). New York: Springer. Retrieved fro m http://books.google.com. Dankiewicz, L.A., Helweg, D.A., Moore, P.W., & Zafran, J.M. (2002). Discrimination of amplitude modulated synthetic echo trains by an echolocating bottlenose dolphin. Journal of the Acoustical Society of America, 112 (4), 1702 1708. doi:10.1121/1.1504856 DeLong, C.M., Au, W.W.L., Harley, H.E., Roitblat, H.L., & Pytka, L. (2007). Human Listeners Provide Insights Into Echo Features Used by Dolphins ( Tursiops truncatus ) to Discriminate Among Objects. Journal of Comparative Psychology, 121 (3), 306 319. doi:10.1037/0735 7036.121.3.306
EFFECTS OF OBJECT RECOGNITION ON ECHOLOCATION 28 DeLong, C.M., Au, W.W.L., Lemonds, D.W., Harley, H.E., & Roitblat, H.L. (2006). Acoustic features of objects matched by an echolocating bottlenose dolphin. Journal of the Acoustical Society of America, 119 (3), 1867 1879. doi:10.1121/1.2161434 Harley, H.E. & DeLong, C.M. (2008) Echoic Object Recognition by the Bottlenose Dolphin. Comparative Cognition & Behavior Reviews, 3 46 65. doi:10.3819/ccbr.2008.30003 Harley, H.E., Fellner, W., & Stamper, M.A. (2010). Cognitive Research with Dolphins ( Tursiops truncatus ) at Disney's The Seas: A Program for Enrichment, Science, Education, and Conservation. International Journal of Comparative Psychology, 23 331 343. Harley, H.E., Putman, E.A., & Roitblat, H.L. (2003). Bottlenose dolphins perceive object features through echolocation. Nature, 424 667 669. Harley, H.E., Roitblat, H.L., & Nachtigall, P.E. (1996). Object representation in the bottlenose dolphin ( Tursiops trun catus ): Integration of visual and echoic information. Journal of Experimental Psychology: Animal Behavior Processes, 22 (2), 164 174. Helweg, D.A., Au, W.W.L., Roitblat, H.L., & Nachtigall, P.E. (1996). Acoustic basis for recognition of aspect depende nt three dimensional targets by an echolocating bottlenose dolphin. Journal of the Acoustical Society of America, 99 (4), 2409 2420. Herman, L. M. & Gordon, J. A. (1974). Auditory delayed matching in the bottlenose dolphin. Journal of the Experimental Analysis of Behavior, 21, (1), 19 26.
EFFECTS OF OBJECT RECOGNITION ON ECHOLOCATION 29 Herman, L.M., Pack, A.A., & Hoffmann Kuhnt, M. (1998). Seeing through sound: Dolphins ( Tursiops truncatus ) perceive the spatial structure of objects through echolocation. Journal of Comparative Psychology, 112 (3), 292 305. Herman, L. M. & Thompson, R. K. R. (1982). Symbolic, identity, and probe delayed matching of sounds by the bottlenose dolphin. Animal Learning and Behavior, 10, (1), 22 34. Herzing, D. L. (1996). Vocalizations and associated underwater behavior of free ranging Atlantic spotted dolphins, Stenella frontalis and bottlenose dolphins, Tursiops truncatus Aquatic Mammals, 22 (2), 61 79. Houser, D.S., Helweg, D.A., & Moore, P.W. ( 1999). Classification of dolphin echolocation clicks by energy and frequency distributions. Journal of the Acoustical Society of America, 106 (3), 1579 1585. Houser, D., Martin, S.W., Bauer, E.J., Phillips, M., Herrin, T., Cross, M., Vidal, A., & Moor e, P.W. (2005). Echolocation characteristics of free swimming bottlenose dolphins during object detection and identification. Journal of the Acoustical Society of America, 117 (4), 2308 2317. Janik, V. M. (2000). Whistle matching in wild bottlenose dolp hins ( Tursiops truncatus ). Science, 289 1355 1357. Janik, V.M. (2009) Acoustic communication in delphinids. Advances in the Study of Behaviors, 40 123 157. doi:10.1016/S0065 3454(09)40004 4 Janik, V. M. & Slater, P. J. (1998). Context specific use sugg ests that bottlenose dolphin signature whistles are cohesion calls. Animal Behavior, 56 829 838.
EFFECTS OF OBJECT RECOGNITION ON ECHOLOCATION 30 Johnson, C. S. (1967). Sound detection thresholds in marine mammals. In W. Tavolga (Ed.), Marine Bioacoustics (pp. 247 260). New York: Pergamon Press. Ret rieved from http://books.google.com. Jolicoeur, P. (1985). The time to name disoriented natural objects. Memory and Cognition 13 (4), 289 303. Lammers, M. O., Au, W. W. L., Aubauer, R., & Nachtigall, P. E. (2003). A comparative analysis of echolocation and burst pulse click trains in Stenella longirostris In Thomas, J., Moss, C., & Vater, M. (Eds.), Echolocation in Bats and Dolphins (pp. 414 419). Chicago, IL: University of Chicago. Lammers, M. O., Au, W. W. L., & Herzing, D. L (2003). The broadband social acoustic signaling behavior of spinner and spotted dolphins. Journal of the Acoustical Society of America, 114 (3), 1629 1639. doi:10.1121/1.1596173 McCowan, B. & Reiss, D. (1995). Whistle contour development in captive bor n infant bottlenose dolphins ( Tursiops truncatus ): Role of learning. Journal of Comparative Psychology, 109 (3), 242 260. Moore, P.W.B., Hall, R.W., Friedl, W.A., & Nachtigall, P.E. (1984). The critical interval in dolphin echolocation: What is it? Jou rnal of the Acoustical Society of America, 76 (1), 314 317. Overstrom, N. A. (1983). Association Between Burst Pulse Sounds and Aggressive Behavior in Captive Atlantic Bottlenosed Dolphins ( Tursiops truncatus ). Zoo Biology, 2 93 103.
EFFECTS OF OBJECT RECOGNITION ON ECHOLOCATION 31 Pack, A.A. & H erman, L.M. (1995) Sensory integration in the bottlenosed dolphin: Immediate recognition of complex shapes across the senses of echolocation and vision. Journal of the Acoustical Society of America, 98 (2), 722 733. Richardson, W. J., Greene, C. R., Mal me, C. I., & Thomson, D. H. (1995). Marine Mammals and Noise Sandiego, California: Academic Press, Inc. Roitblat, H.L. (2002). The cognitive dolphin. In Bekoff, M., Allen, C., & Burghardt, G. (Eds.), The Cognitive Animal: Empirical and Theoretical Pers pectives on Animal Cognition (pp. 183 187). Cambridge, MA: The MIT Press. Roitblat, H.L., Penner, R.H., & Nachtigall, P.E. (1990). Matching to Sample by an Echolocating Dolphin ( Tursiops truncatus ). Journal of Experimental Psychology: Animal Behavior Processes, 16 (1), 85 95. Sayigh, L. S., Tyack, P. L., Wells, R. S., & Scott, M. D. (1990). Signature whistles of free ranging bottlenose dolphins Tursiops truncatus : Stability and mother offspring comparisons. Behavioral Ecology and Sociobiology, 26 ( 4), 247 260. Shinar, D. & Owen, D. H. (1973). Effects of form rotation on the speed of classification: The development of shape constancy. Perception and Psychophysics, 14, (1), 149 154. Thompson, R. K. R. & Herman, L. M. (1975). Underwater frequency di scrimination in the bottlenosed dolphin (1 140 kHz) and the human (1 8 kHz). Journal of the Acoustical Society of America, 57 (4), 943 948. United States Navy (n.d.). Marine Mammal Program. Retrieved from http://www.public.navy.mil/spawar/Pacific/7150 0/Pages/default.aspx.
EFFECTS OF OBJECT RECOGNITION ON ECHOLOCATION 32 Watwood, S. L., Owen, E. C. G., Tyack, P. L., & Wells, R. S. (2005). Signature whistles used by temporarily restrained and free swimming bottlenose dolphins, Tursiops truncatus Animal Behavior, 69 1373 1386. doi:10.1016/j.anbehav.2004.08.019 Watwood, S. L., Tyack, P. L., & Wells, R. S. (2004). Whistle sharing in paired male bottlenose dolphins, Tursiops truncatus Behavioral Ecology and Sociobiology, 55 531 543. doi:10.1007/s00265 003 0724 y Xitco, Jr., M.J & Roitblat, H.L. (1996). Object recognition through eavesdropping: Passive echolocation in bottlenose dolphins. Animal Learning & Behavior, 24 (4), 355 365. Wood, F. G. (1953). Underwater sound production and concurrent behavior of captive porpoises, Tursiops truncatus and Stenella plagiodon Bulletin of Marine Science, 3 (2), 120 133.
EFFECTS OF OBJECT RECOGNITION ON ECHOLOCATION 33 Table 1 Performance Accuracy Across All Sessions for All Object Sets. Analyzed Sessions Are Highlighted in G rey. Object Set Session Accuracy 11% Poor (May 2009) 1 33.33% 2 33.33% 3 44.44% 4 33.33% 5 44.44% 28% Poor (January 2010) 1 38.89% 2 38.89% 3 22.22% 4 27.78% 5 50% 11% Good (October 2010) 1 94.44% 2 88.89% 3 94.44% 4 100% 5 100% 28% Good (April 2009) 1 50% 2 55.56% 3 61.11% 4 50% 5 77.78%
EFFECTS OF OBJECT RECOGNITION ON ECHOLOCATION 34 Table 2 Accuracy, Average Number of Clicks, Average Time Echolocating, Average Number of Click Trains, and Occurrence of the Terminal Burst Pulse for Each Object Set. Object Set Session Accuracy Number of Clicks Emitted Time Spent Echolocating (seconds) Number of Click Trains Terminal Burst (percent of trials) 11% Poor 1 33.33% 188.89 3.63 2.11 66.67% 5 44.44% 240.44 4.76 2.28 83.33% 28% Poor 1 38.89% 243.61 5.16 1.61 83.33% 3 22.22% 230.83 6.84 2.11 55.56% 5 50% 233 5.56 2.25 81.25% 11% Good 1 94.44% 207.94 5.42 2 100% 2 88.89% 192.53 5.68 2.59 94.12% 5 100% 160.39 4.8 2.22 88.89% 28% Good 1 50% 221.61 4.84 2.33 55.56% 5 77.78% 165.22 3.81 1.83 38.89%
EFFECTS OF OBJECT RECOGNITION ON ECHOLOCATION 35 Figure 1: Dolphin Vocalizations Figure 1 Spectrograms, as a function of frequency over time, of the three categories of dolphin vocalizations. From left to right: whistle with the fundamental frequency around 8kHz and two harmonics above; burst pulse; echolocation clicks.
EFFECTS OF OBJECT RECOGNITION ON ECHOLOCATION 36 Figure 2: The Facility Figure 2 The facility included three pools: Main Tank, A Pool, and B Pool. The current study occurred within B Pool. !"#$$% &'()*+'), -"#$$% AG a t e BG a t e (S p yball) !"#$%"# &'()*+ ,-..*/,)#.0 !)*)( Y ell o w P l a t f o r m 123445$ 67899.67899.= Apo ol (ch1) = Bpo ol (ch2) = Inside B-g a t e/B-g a t e = MT-B = Inside A-g a t e/A-g a t e = MT-A ./001 0 2 3 4 5 1 2 3 4 5 6 = Apo ol = Bp ool = C anal = B-g a t e = B-s k im #67"./00 0 8 9 : 7 8 9 7 m 8.2 m Depth= 2.1 m Hydrophone
EFFECTS OF OBJECT RECOGNITION ON ECHOLOCATION 37 Figure 3: The Object Sets Figure 3 The objects within each object set. Objects in Object Set 11% Poor: saxophone, guitar, and trumpet. Objects in Object Set 28% Poor: jellyfish, sea turtle, and whale. Objects in Object Set 11% Good: 3 way, spoon, and tube. Objects in Object Set 28% Good: sprinkler, watering can, and nozzle.
EFFECTS OF OBJECT RECOGNITION ON ECHOLOCATION 38 Figure 4 : The Task Set Up Figur e 4 The task set up within B Pool. The dolphin swam 6.1 m from the trainer to the sample, then another 6.1 m to the alternatives. Objects were located 0.7 m from the walls of the tank with 1.2m between each alternative, and were immersed in the water such that the center of each object was 40.6 cm below the surface of the water. !"#$$% Trainer Sample & Hydrophone Alt 3 Alt 2 Alt 1 6.1 m 6.1 m 0.7 m 0.7 m 0.7 m 0.7 m Research Assistant 1.2 m 1.2 m
EFFECTS OF OBJECT RECOGNITION ON ECHOLOCATION 39 Figure 5: Click Counting Figure 5 Once a click train is selected (top image, x axis is time and y axis is frequency), the number of clicks can be counted in Avisoft's Pulse Train Analysis feature. In the bottom left image, hysterisis is the amount of energy by which the click must differ from the background noise to be recognized, and the threshold is the minimum energy each click must have t o be recognized. The bottom right image shows the clicks in the Pulse Train Analysis (x axis is time and y axis is energy), where the black horizontal line towards the bottom is threshold.
EFFECTS OF OBJECT RECOGNITION ON ECHOLOCATION 40 Figure 6: Average Number of Clicks With Respect to Familiarity an d Accuracy Figure 6 The average number of clicks per session per object set. Accuracy of each session is also shown via the blue line and object familiarity is represented as sessions 1 (unfamiliar) and 5 (familiar) !" #!" $!!" $#!" %!!" %#!" &!!" !'" $!'" %!'" &!'" (!'" #!'" )!'" *!'" +!'" ,!'" $!!'" $" #" $" &" #" $" %" #" $" #" $$'"-../" %+'"-../" $$'"0..1" %+'"0..1" !"#$%'(')*'+,-./0'1#$'2#00-)3' 4#$.#35'+)$$#.5' 2"345678" 966:/;6<"
EFFECTS OF OBJECT RECOGNITION ON ECHOLOCATION 41 Figure 7: Time Spent Echolocating With Respect to Familiarity and Accuracy Figure 7 Average time spent echolocating the sam ple per session per object set. Accuracy of each session is shown via the blue line and object familiarity is repres ented as sessions 1 (unfamiliar) a nd 5 (familiar) !" $" %" &" (" #" )" *" +" !'" $!'" %!'" &!'" (!'" #!'" )!'" *!'" +!'" ,!'" $!!'" $" #" $" &" #" $" %" #" $" #" $$'"-../" %+'"-../" $$'"0..1" %+'"0..1" !"#$%'6-7#'21#35'8.9),).%5-3&' 1#$'2#00-)3':0#.)3;0<' 4#$.#35'+)$$#.5' =5>?"345675@A" 966:/;6<"
EFFECTS OF OBJECT RECOGNITION ON ECHOLOCATION 42 Figure 8: Number of Click Trains With Respect to Familiarity and Accuracy Figure 8 The average number of click trains emitted by the dolphin across object sets and sessions. Accuracy of each session is shown via the blue line, and object familiarity is represented as sessions 1 (unfamiliar) and 5 (familiar). !" !B#" $" $B#" %" %B#" &" !'" $!'" %!'" &!'" (!'" #!'" )!'" *!'" +!'" ,!'" $!!'" $" #" $" &" #" $" %" #" $" #" $$'"-../" %+'"-../" $$'"0..1" %+'"0..1" !"#$%'(')*'8.9)'6$%-30'1#$'2#00-)3' 4#$.#35'+)$$#.5' 2"C6D."=/;5@8" 966:/;6<"
EFFECTS OF OBJECT RECOGNITION ON ECHOLOCATION 43 Figure 9: Occurrence of the Terminal Burst With Respect to Familiarity and Accuracy Figure 9 The percentage of trials within each session of each object set in which the dolphin emitted a terminal burst pulse after echolocating the sample. Accuracy of each session is shown via the blue line, and object familiarity is represented as sessions 1 (unf amiliar) and 5 (familiar). !'" $!'" %!'" &!'" (!'" #!'" )!'" *!'" +!'" ,!'" $!!'" !'" $!'" %!'" &!'" (!'" #!'" )!'" *!'" +!'" ,!'" $!!'" $" #" $" &" #" $" %" #" $" #" $$'"-../" %+'"-../" $$'"0..1" %+'"0..1" 4#$.#35')*'6$-%,0'=-59'6#$7-3%,' >?$05' 4#$.#35'+)$$#.5' E66://?@6?".F"=?/>5@;4"G:/8H" 966:/;6<"