Caroline M. DeLong
EDUCATION	TEACHING	RESEARCH	PAPERS	RAs	LINKS

Comparative Cognition and Sensory Perception Research

The goal of my research has been to elucidate the information-bearing parameters of dolphin echolocation and communication signals. Determining these parameters will help reveal the nature of dolphins' cognitive representations.

Acoustic Rhythm Discrimination in the Bottlenose Dolphin
Phantom Echo Projects (Dolphin Echolocation)
Dissertation Research: Echoic Cues that Permit Cross-Modal Matching in the Bottlenose Dolphin

Acoustic Rhythm Discrimination in the Bottlenose Dolphin

Previous research has shown that dolphins attend to the frequency, duration, and loudness of sounds. The purpose of this research was to determine whether dolphins perceive variations in temporal patterns (i.e., rhythm) in a series of sounds. Both studies were conducted at Epcot's Living Seas in Orlando, Florida. In the first study, a 2-year-old female Atlantic bottlenose dolphin (Tursiops truncatus) named Naia was presented with six different acoustic stimuli (DeLong, 1997). Stimuli included five regular, rhythmic patterns and one irregular, arrhythmic pattern. The rhythms varied in the duration of the tone and intertone intervals (the frequency and loudness of the sounds was held constant over all the patterns). For example, one rhythm ("linear") consisted of a 250 msec. tone that was alternated with a 250 msec. intertone interval (silence) and played continuously. Naia's ability to discriminate among the different temporal patterns was tested by asking her to respond differentially to each of the patterns. For example, in response to pattern #3, she was rewarded for performing an aerial leap, whereas in response to pattern #2, she was rewarded for doing a spin. At the end of this first study, the dolphin was not responding differentially to the rhythms. This was apparently due to the inexperience of the subject and abbreviated training sessions.

This research was extended in a second study in which an adult male Atlantic Bottlenose dolphin was presented with six temporal patterns (Harley, Odell, Putman, Goonen, & DeLong, 2002). The dolphin was rewarded for performing a different behavior to each of the six rhythms, e.g., a spin to Rhythm 1, ball toss to Rhythm 2, etc. The dolphin responded differentially to the rhythms. Performance accuracy over the final ten 18-trial sessions averaged 93%. In transfer tests, the rhythms were altered either in frequency or in tempo. When the rhythms were altered in frequency, the dolphin continued to perform well. When rhythms were slightly slowed down or speeded up, performance remained high, but accuracy was lower with more extreme tempo changes.

This research suggests that dolphins perceive and represent the temporal pattern (i.e., rhythm) when they hear a series of sounds.
This may have important implications for the way dolphins process whistles (communication signals). The temporal pattern of sounds may carry information, such as the species or individual identification of the caller.

Phantom Echo Projects (Dolphin Echolocation)

Dolphins echolocate by emitting clicks, which travel through the water, bounce off objects, and return to the dolphin as echoes. The echoes dolphins receive are often complex, varying along several parameters such as amplitude, frequency bandwidth, and highlight structure. It is not clear how these echo parameters interact to convey information about the object to the dolphin (e.g., the shape or material of the object). These studies involved the use of new technology, the Phantom Echo Generator, which allows experimenters to produce a “phantom” echo which is an acoustically simulated phantom replica of an object echo (it does not originate from a real object). This phantom technique permits the experimenter to extract and control echo parameters separately, which allows for a better understanding of what each echo parameter contributes to the dolphin’s representation of the object.

PEG setup

Experimental configuration for the Phantom Echo Experiments

In one study, the capability of an echolocating dolphin to discriminate between acoustically phantom replicas of targets and their real equivalents was tested (Aubauer, Au, Nachtigall, Pawloski, & DeLong, 2000). A female Atlantic bottlenose dolphin named BJ was trained to respond to a target presentation in a go/no-go procedure. BJ's task was to provide one behavior (the "go", which was backing out of a hoop station) in response to one class of targets (a steel sphere was designated as the standard target), and a different behavior (the "nogo", which is remaining in the hoop station) in response to another class of targets (the comparison targets included all other targets; brass, aluminum, nylon spheres). BJ was successful in classifying these targets using echolocation. Probe trials were inserted in which BJ was presented with "phantom" steel echoes instead of receiving echoes from the real steel target. BJ classified the phantom steel echoes in the same manner that she classified the real steel echoes. These results suggest that it seems as though dolphins will accept phantom echoes as coming from real targets.

BJ during Phantom experiment

BJ during Phantom Echo Experiments

I conducted a study to examine the characteristics of the dolphin's outgoing sonar signals (primarily the quantity of signals) during echolocation of real and phantom targets (DeLong, Au, Nachtigall, Aubauer, & Roitblat, 1999). The echolocation task is described above. I found that the dolphin emitted an average of about 70 clicks per trial during both tasks (accuracy with 88-93% correct). Significantly more clicks were emitted on trials ending in an incorrect response than on those ending in a correct choice. In addition, the dolphin emitted fewer clicks when echolocating certain target types. Examining the phantom target trials revealed that she emitted the same number of clicks when echolocating real steel vs. phantom steel targets, but the distribution of click quantities differed significantly for phantom vs. real steel targets. The dolphin's use of a large number of clicks and high level of accuracy suggest that the dolphin combines information from successive echoes until a confidence level is reached, rather than basing her decision on individual clicks. These results also suggest that examining the dolphin's outgoing sonar signals may reveal the use of different cognitive strategies.

These studies suggest that dolphins accept phantom echoes as coming from real objects, and that examining their outgoing signals can provide insight into their cognitive processes. The next step is to systematically test echo parameters to assess the type of information each one conveys to the dolphin.

Dissertation Research: Echoic Cues that Permit Cross-Modal Matching in the Bottlenose Dolphin

My dissertation research has two goals: (1) to ascertain the echoic features bottlenose dolphins use to determine the structural properties of objects (e.g., size, shape, and material composition), and (2) to clarify the nature of the representations used by dolphins during cross-modal matching tasks. There are three parts to the study: a series of cross-modal matching experiments with a bottlenose dolphin subject , acoustic measurements of the objects used in the task (collection of object echoes), and a human listening experiment in which human participants are asked to classify the object echoes. This research takes place at two sites: Epcot's Living Seas in Orlando, and the Hawaii Institute of Marine Biology in Kaneohe, Oahu, Hawaii.

Toby

Toby at Epcot

In Part 1, a male Atlantic bottlenose dolphin, Toby, was presented with a series of cross-modal matching tasks. Toby was presented with a sample object in one modality (vision or echolocation), and then asked to choose the same object from among three comparison objects using a different modality (echolocation or vision, respectively). In the first experiment, the objects were multidimensional, i.e., they varied along several object dimensions such as size, shape, material, and texture. In the second experiment, the objects were unidimensional, i.e., they varied along primarily one dimension. One group of objects varied only in size, while other groups of objects varied only in shape, or material, or texture. Toby was able to cross-modally match many of the objects, but his performance varied between object sets.

echo collection

Collecting the echoes from a target. The binaural hydrophone in on the right, and the target is suspended from the rotor on the left.

In Part 2, I made acoustic measurements of the objects used in Part 1 so that I could examine the echoes that dolphin received during the cross-modal matching task. Acoustic measurements were made by projecting a dolphin-like click at each object as it hung suspended in a pool of water. Objects were measured from many angles, to simulate the variation in echoes dolphin would receive as they inspected the object from different orientations. Echoic features (e.g., target strength, duration) were extracted from each echo. The echoic features that vary across objects can be examined in conjunction with the dolphin’s performance in Part 1. This analysis can reveal the parts of the echoes dolphins use to determine the features of objects. For example, if the dolphin confused two objects of different shapes that had very similar echo highlight structures, then it suggests that echo highlight structure is one feature that dolphins use to determine the shape of objects.

In Part 3, human participants will listen to the object echoes. The participants will be asked to discriminate among the objects (similar to the task given to the dolphin), and then asked to report the cues in the echoes that made it possible to discriminate among the echoes (e.g., duration or loudness of echo). It is sensible to hypothesize that the cues used by the human listeners could also be used by dolphins because the inner ear of dolphins functions similarly to the human inner ear, but in a higher frequency range. Thus, the results of the human listening study should provide insight into the echoic features dolphins use to determine the structural properties of objects.

Funding for this dissertation research was provided by:

American Association for University Women Honolulu Branch - Pacific Dissertation Fellowship
Soroptimist International - Founder Region Dissertation Fellowship
SEASPACE Organization