Visual Object Recognition in North American River Otters (Lontra canadensis) in Collaboration with the Seneca Park Zoo in Rochester, NY

    The goal of the project is to investigate visual object recognition in North American river otters. The focus of this study is to explore visual features of objects that are important to river otters.  The otters are being trained to perform an match-to-sample task and then will be tested using a variety of objects. Different object sets will vary in color, size, and shape.  This study takes place at the Seneca Park Zoo.

Collaborator:  Catina Wright (Seneca Park Zoo)

This research will be featured in the RIT Athenaeum in the August, 2012 issue

Numerosity Discrimination in Goldfish

The goal of this study is to investigate the ability of goldfish (Carassius auratus) to use number to compare two groups of objects. Two goldfish were trained individually in five-gallon test tanks. First the fish were trained to eat crumble food from a training rod in the test tank. Then the fish were trained to swim through one of two 6.5 cm holes in a piece of corrugated plastic to receive the food reward. Currently, the fish are being trained to swim through the hole with the positive stimulus above it. The stimuli consist of laminated white paper with black 2D shapes (circles, rectangles, triangles). The first pair of stimuli being tested is the ratio 1:2 (the positive stimulus is 1 for one fish and 2 for the other fish).  Once the fish can successfully discriminate between 1 and 2, other ratios will be tested (e.g., 1:3, 2:3). All stimuli pairs are matched for surface area and contain similar shapes so that the fish attend to the number of stimuli. Throughout the experiment, the individual fish must consistently swim through the hole with either the lower or higher numerosity above it, depending on how they were trained for the first ratio. The results of this study can be compared to research on numerosity discrimination in chimpanzees, monkeys, dolphins, and mosquitofish. 

Shape Discrimination in Goldfish

    The goal of this study is to investigate 2D visual object recognition in goldfish (Carassius auratus). We want to determine visual features goldfish use to discriminate among objects (e.g., length, width, surface area, and diameter). Three goldfish were trained individually in 5 gallon test tanks to eat a mixture of flake food and water from a 1.0 ml syringe. Then the fish were trained to tap a black circle with a 2 cm diameter on a white background to receive the food reward by pressing their mouth to the circle. Stimuli were printed on a laminated piece of paper attached to a piece of corrugated plastic, which was presented underwater. The fish were able to reliably tap the circle within 11-15 sessions. In the first object discrimination test, a 3.1 cm x 1 cm black rectangle (matched for surface area) was presented alongside the circle. The fish successfully chose the circle across 77 sessions (M = 83%).  In the second object discrimination test, a 2 cm x 1 cm black rectangle (matched for length) was presented alongside the circle. The average choice accuracy was 76% after the second test.  These results are similar to those reported for reef fish. More tests are underway with different shape pairs.

This research was featured in the RIT Athenaeum in the October, 2011 issue here:

Also see the youtube video here:

Can Dolphins Recognize a Novel Object From Any Orientation? A Comparative Study of Object Constancy in Dolphins and Humans

    Every day we quickly and accurately recognize objects visually regardless of changes in object orientation.  This ability to recognize an object from any viewpoint is called object constancy and it is one of the fundamental and essential properties of human visual perception.  While humans rely heavily on vision to identify objects in their environment, dolphins use echolocation (i.e., biological sonar).  Dolphins echolocate by emitting a series of ultrasonic clicks and listening to the returning echoes. Dolphins use echolocation to navigate, avoid predators, and track moving prey. Most of the objects dolphin encounter are aspect-dependent; meaning the size and shape of the surfaces of the object will change as they are viewed from different orientations. The echoes from these types of object can vary considerably depending on the angle from which they are inspected by the dolphin.  In fact, echoes from different orientations of a single object can vary more from each other than do echoes from different objects. Thus, the problem echolocating dolphins face in recognizing objects is very similar to the one humans face – the object must be correctly identified despite large changes in the specific sensory information (the visual image or the auditory event) that result from changes in the orientation of the object.


Collaborators:  Heidi Harley, Whitlow Au

Current Projects

Discrimination of Fish Prey by Human Listeners using Dolphin and Porpoise Echolocation Signals

        Dolphins and porpoises can detect and discriminate among objects using echolocation.  Dolphins use a short broadband signal, whereas porpoises use a longer, more narrow-band signal with a lower source level. The goal of this study is to explore the different echoic cues that may be available to dolphins vs. porpoises during a discrimination task as a result of their different signals.  A recent and fruitful approach for identifying potential salient cues is to present object echoes to human listeners, who can typically perform as well as dolphins on object discrimination tasks.  Echoes were obtained from four fish species - sea bass (Dicentrarchus labras), Pollack (Pollachius pollachius), grey mullet (Chelon labrosus), and Atlantic cod (Gadus morhua) - using both broadband bottlenose dolphin and narrow-band harbor porpoise signals.  These echoes were collected along the lateral axis of the each fish (including head, broadside, and tail aspects).  Human listeners will be asked to discriminate among the fish using first dolphin echoes, then porpoise echoes (and vice versa), as well as report the echoic features that they used.  The performance of the human listeners can elucidate potential differences in how dolphins and porpoises make use of acoustic features of echoes to identify fishes.

Acoustic Features of Objects Matched by an Echolocating Bottlenose Dolphin

        The focus of this study was to investigate how dolphins use acoustic features in returning echolocation signals to discriminate among objects.  An echolocating dolphin performed a match-to-sample task with objects that varied in size, shape, material, and texture. After the task was completed, the features of the object echoes were measured (e.g., target strength, peak frequency).  The dolphin’s error patterns were examined in conjunction with the between-object variation in acoustic features to identify the acoustic features that the dolphin used to discriminate among the objects.  This study explored two hypotheses regarding the way dolphins use acoustic information in echoes: (1) use of a single feature, or (2) use of a linear combination of multiple features.  The results suggested that dolphins do not use a single feature across all object sets or a linear combination of six echo features.  Five features appeared to be important to the dolphin on four or more sets: the echo spectrum shape, the pattern of changes in target strength and number of highlights as a function of object orientation, and peak and center frequency.  These data suggest that dolphins use multiple features and integrate information across echoes from a range of object orientations. 

Human Listeners Provide Insights into Echo Features Used by Dolphins to Discriminate Among Objects 

        Echolocating bottlenose dolphins (Tursiops truncatus) discriminate between objects on the basis of the echoes reflected by the objects.  However, it is not clear which echo features are important for object discrimination.  To gain insight into the salient features, a dolphin performed a match-to-sample task and then human listeners were presented via headphones with echoes from the same objects used in the dolphin’s task.  In two experiments, human listeners performed as well or better than the dolphin at discriminating objects and reported the salient acoustic cues.  The error patterns of the humans and the dolphin were compared to discern which acoustic features were likely to have been used by the dolphin. The results indicated that the dolphin did not appear to use overall echo amplitude, but that it attended to the pattern of changes in the echoes across different object orientations.  Human listeners can quickly identify salient combinations of echo features that permit object discrimination, which can be used to generate hypotheses that can be tested using dolphins as subjects.

Echo Features Used by Human Listeners to Discriminate Among Objects that Vary in Material or Wall Thickness: Implications for Echolocating Dolphins

        Echolocating dolphins extract object feature information from the acoustic parameters of echoes.  To gain insight into which acoustic parameters are important for object discrimination, human listeners were presented with echoes from objects used in two discrimination tasks performed by dolphins: hollow cylinders with varying wall thicknesses (± 0.2, 0.3, 0.4, and 0.8 mm), and spheres made of different materials (steel, aluminum, brass, nylon, and glass).  The human listeners performed as well or better than the dolphins at the task of discriminating between the standard object and the comparison objects on both the cylinders (humans = 97.1%; dolphin = 82.3%) and the spheres (humans = 86.6%; dolphin = 88.7%).  The human listeners reported using primarily pitch and duration to discriminate among the cylinders, and pitch and timbre to discriminate among the spheres.  Dolphins may use some of the same echo features as the humans to discriminate among objects varying in material or structure.  Human listening studies can be used to quickly identify salient combinations of echo features that permit object discrimination, which can then be used to generate hypotheses that can be tested using dolphins as subjects.

Evidence for Spatial Representation of Object Shape by Echolocating Bats (Eptesicus fuscus)

        Big brown bats were trained in a two-choice task to locate a 2-cylinder dipole object with constant 5-cm spacing in the presence of either a 1-cylinder monopole or another 2-cylinder dipole with shorter spacing.  For the dipole vs monopole task, the objects were either stationary or in motion during each trial.  The dipole and monopole objects varied from trial to trial in left-right position while also roving in range (15-40 cm), crossrange separation (10-40 cm), and dipole aspect angle (0°-90°).  These manipulations prevented any single feature of the acoustic stimuli from being a stable indicator of which object was the correct choice.  After accounting for effects of masking between echoes from pairs of cylinders at similar distances, the bats discriminated the 5-cm dipole from both the monopole and dipole alternatives with performance independent of aspect angle, implying a distal, spatial object representation rather than a proximal, acoustic object representation.  

Past Projects

Mal, a bottlenose dolphin, waiting to listen to echoes (photo by Wendi Fellner).

A big brown bat (Eptesicus fuscus) in an object discrimination task (photo by C. DeLong).

Research assistant Irene Kannyo formatting echo stimuli for a human listening study (photo by C. DeLong).

Heather the river otter (photo by A. Sue Weisler/RIT).

Dorado the goldfish swims through the hole with two stimuli in the training phase for the numerosity study (photo by C. DeLong).

Poseidon the goldfish has to tap the circle to receive a food reward in the shape discrimination study (photo by C. DeLong).

BJ, a bottlenose dolphin, waiting for a test trial to begin in an echolocation task (photo by C. DeLong).

Harbor porpoises (photo by Bill Curtsinger).