ABSTRACTS

Spatial generalization and peak shift in humans.

Cheng, Ken; Spetch, Marcia L.

Source
Learning & Motivation. Vol 33(3) Aug 2002, 358-389.
Academic Press, US

Abstract
Using a computer betting game, five experiments tested male and female university students (aged 17-44 yrs) on spatial generalization and peak shift. On each trial, one location was marked and the subject was invited to bet 0-4 points. At the winning location (S+), bets won four times the points betted. At nearby losing locations (S-s), points betted were lost. Generalization gradients were exponential in shape, supporting R. N. Shepard's (1987), law (Experiment 1). With peak shift manipulations, three kinds of peak shift or area shift were found. Subjects betted more on the S+ side than on the S- side (Experiments 2-4). When asked if a location was the winning location, subjects responded "yes' more often to locations on the S+ side than to locations on the S- side (Experiments 3-5). When asked to point to the winning location on the screen, subjects' errors indicated peak shift (Experiment 5).

The effect of distinctive parts on recognition of depth-rotated objects by pigeons (Columba livia) and humans.

Spetch, Marcia L; Friedman, Alinda; Reid, Sheri L.

Source
Journal of Experimental Psychology: General. Vol 130(2) Jun 2001, 238-255.
American Psychological Assn, US

Abstract
To explore whether effects observed in human object recognition represent fundamental properties of visual perception that are general across species, the authors trained pigeons (Columba livia) and humans to discriminate between pictures of 3-dimensional objects that differed in shape. Novel pictures of the depth-rotated objects were then tested for recognition. Across conditions, the object pairs contained either 0, 1, 3, or 5 distinctive pails. Pigeons showed viewpoint dependence in all object-part conditions, and their performance declined systematically with degree of rotation from the nearest training view. Humans showed viewpoint invariance for novel rotations between the training views but viewpoint dependence for novel rotations outside the training views. For humans, but not pigeons, viewpoint dependence was weakest in the 1-part condition. The authors discuss the results in terms of structural and multiple-view models of object recognition.

Blocking in landmark-based search in honeybees.

Perception of pictorial depth cues by pigeons

Chapter: Mechanisms of landmark use in mammals and birds.

Contiguity and Conditioned Reinforcement in Probabilistic Choice

Encoding of spatial information in images of an outdoor scene by pigeons and humans.

Marcia L. Spetch, Debbie M. Kelly and David P. Lechelt

University of Alberta

Pigeons and adult humans searched for a 2 cm square unmarked goal in digitized images of an outdoor scene presented on a touch-screen monitor. In Experiment 1, the scene contained three landmarks near the goal and a visually rich background. Six training images presented the scene from different viewing directions and distances. Subsequent unreinforced tests in which landmark or background cues were removed or shifted revealed that pigeons' search was controlled by both proximal landmarks and background cues, whereas humans relied only on the proximal landmarks. Pigeons's search accuracy dropped substantially when they were presented with novel views of the same scene, whereas humans showed perfect transfer to novel views. In Experiment 2, pigeons with previous outdoor experience and humans were trained with 28 views of an outdoor scene. Both pigeons and humans transferred well to novel views of the scene. This positive transfer suggests that under some conditions pigeons, like humans, may encode the three-dimensional spatial information in images of a scene

Spatial Peak Shift and Generalization in Pigeons

Ken Cheng

Macquarie University

Marcia L. Spetch and Michael Johnston

University of Alberta

These experiments examined how pigeons generalize across spatial locations. During training, a square was presented at a fixed height at one of two horizontal locations on a monitor screen. One location (S+) signaled reward while the other one (S-) signaled no reward. The birds were then tested occasionally with a range of locations. After training with S+ only, the generalization gradient peaked at S+ and was approximately Gaussian in shape. After training with equal numbers of S+ and S- trials, response rates were higher on the S+ side of the distribution. This asymmetry diminishing over testing. When the S+ and S- were close together, the peak of responding was shifted on initial generalization tests. Generalization gradients along the orthogonal vertical dimension were approximately exponential in shape. This is the first demonstration of generalization and peak shift in the spatial domain.

Pigeons' Use of Landmarks for Spatial Search in a Laboratory Arena and in Digitized Images of the Arena.

David P. Lechelt and Marcia L. Spetch

University of Alberta

Pigeons were trained in a touch-screen task and in an open-field task to search for a hidden goal using an array of landmarks to guide their search behavior. The touch-screen task presented digitized images of the open field landmarks and environment. Upon completion of training in each task, tests involving the alteration, shifting, or removal of landmarks were conducted to determine which landmark(s) and stimulus feature(s) controlled search behavior. In both tasks, proximity to the goal was an important determinant of landmark control. The overall patterns of landmark control showed both similarities and differences in the two tasks. No evidence was obtained that learning about the arrangement of landmarks and goal transferred across the two tasks.

Mechanisms of landmark use in mammals and birds.

Ken Cheng, School of Behavioural Sciences Macquarie University

Marcia L. Spetch, Department of Psychology University of Alberta

We have reviewed the transformational approach to the study of landmark-based spatial memory, and extant data on three themes: the use of metric properties of distances and directions, the use of multiple cues, and the use of configurations of landmarks. In the transformational approach, the participant has the task of locating a target with respect to an experimentally provided set of landmarks. Often the experimental set is moved about from trial to trial to ensure that the target is in a constant location only with respect to the experimental landmarks, and not to anything else. On crucial unrewarded tests, the experimental landmarks are transformed in theoretically motivated ways to find out something about how the participants use landmarks in localisation.

Our review indicates that all vertebrate species that have been studied use metric properties in localisation. On transformations of landmark arrays, participants search in locations that preserve some metric properties, although different species might preserve different properties. In pigeons, evidence indicates that distances and directions are computed separately and independently.

With multiple landmarks in the array, participants often use more than one cue or element in searching. However, when other cues provide a stable directional frame of reference, control of searching is sometimes restricted to a single element of a landmark array. We reviewed parallels between landmark learning and other forms of learning in general. Participants prefer to rely on better predictors, those landmarks that are closer to the goal. Moreover, relative distances from goal to landmarks matter. A landmark at a particular absolute distance is overshadowed by a landmark that is nearer to the goal. Testing for other general laws of learning in landmark-based search forms a worthwhile research program.

By the use of configuration, we mean more than using multiple landmarks. The participant must encode some relationship between landmarks. An elemental solution encodes only relations between the target and single landmarks. Evidence indicates that all vertebrate species use the configuration of landmarks, whereas honeybees seem to use an elemental system. Adult humans differ from other species in the use of configuration. They use abstract geometric rules such as "search in the middle of the array". The entire area of the use of configuration deserves more investigation.

Pigeons' Coding of Event Duration in Delayed Matching-To-Sample

Douglas S. Grant, Marcia L. Spetch, and Ronald Kelly

University of Alberta

A fundamental issue in the analysis of working memory is the nature of the representation, or code, that mediates accurate performance across a retention interval. In this chapter we addressed the nature of working memory representations in DMTS tasks in which the samples differ in duration. In the standard choice version of the DMTS task, in which each duration sample is mapped in a one-to-one relation to a comparison stimulus, naive pigeons code samples retrospectively and analogically. We maintain that, in such instances, pigeons represent different event durations as codes which specify values along a dimension that changes in a continuous and cumulative manner as a function of the passage of time. Such a coding process is analogical because the representative dimension shares with duration the property of changing in a continuous and cumulative manner. As one example of an analogical-coding process, pigeons might represent different durations as different intensities of a visual image. On this view, the intensity of the image increases as time spent in the presence of the sample increases, and decreases as time since termination of the sample increases. The phenomena considered in the first major section of this chapter encourage the view that durations are coded analogically, and discourage the view that durations are coded asymmetrically or categorically (either retrospectively or prospectively).

Research considered in the second major section of this chapter suggests, however, that duration samples are not always coded analogically. In the successive DMTS task, in which accuracy is assessed by a go/no-go response rather than by choice behavior, duration samples are coded nonanalogically, perhaps in an instructional, prospective manner. Evidence of nonanalogical coding of duration samples has also been obtained in two instances in the choice DMTS task. First, pigeons initially trained in successive DMTS demonstrate nonanalogical coding when subsequently transferred to a choice DMTS task in which the contingencies are consistent with those of prior training. Substantial positive transfer from successive to choice DMTS suggests that pigeons continue to use the nonanalogical coding process adopted in the successive task when transferred to the choice task. Second, pigeons trained with two sets of samples, either both temporal or one temporal and one nontemporal, mapped onto a single set of comparisons (an MTO mapping) also demonstrate nonanalogical coding of the temporal samples. The results of mediated transfer testing suggest that samples associated with the same comparison stimulus activate a common code (e.g., "peck red", "event A"). One caveat to this conclusion is that an MTO mapping is less likely to reduce the tendency to code temporal samples analogically if accurate matching with temporal samples is acquired prior to acquiring accurate matching with nontemporal samples.

In summary, substantial progress has been made over the past decade in identifying the coding processes used by pigeons in DMTS tasks with duration samples, and in determining some of the factors that induce analogical or nonanalogical forms of coding. As yet, however, less is known about the generality of these results across species (e.g., Santi, Weise & Kuiper, 1995) or the coding processes used in other types of tasks that may require retention of temporal information.

Use of landmark configuration in pigeons and humans: II. Generality across search tasks

Marcia L. Spetch, University of Alberta

Ken Cheng, University of Toronto

Suzanne E. MacDonald, York University

Brie A. Linkenhoker, Transylvania University

Debbie M. Kelly and Sharon R. Doerkson, University of Alberta

Pigeons and humans searched for a goal that was hidden in varied locations within a search space. The goal location was fixed relative to an array of identical landmarks. Pigeons searched on the laboratory floor and humans searched on a table top, or an outdoor field. In Experiment 1, the goal was centered in a square array of 4 landmarks. When the spacing between landmarks was increased, humans searched in the middle of the expanded array, whereas pigeons searched in locations that preserved distance and direction to an individual landmark. In Experiment 2, the goal was centered between and a perpendicular distance away from 2 landmarks aligned in the left/right dimension. When landmark spacing was increased, humans, but not pigeons, shifted their searching away from the landmarks along the perpendicular axis. These results parallel those obtained in touch-screen tasks. Pigeons and humans differ in how they use landmark configuration.

Learning the configuration of a landmark array, I: Touch-screen studies with pigeons and humans

Marcia L. Spetch

University of Alberta

Ken Cheng

University of Toronto

Suzanne E. MacDonald

York University

Pigeons and humans searched on a touch-screen monitor for an unmarked goal located relative to an array of landmarks presented in varied screen locations. After training with the goal centered in various square arrays of four landmarks, humans, but not pigeons, transferred accurately to arrays with novel elements. Humans searched in the middle of expanded arrays, whereas pigeons preserved the distance and direction to a single landmark. When trained with the goal centered below two identical horizontally aligned landmarks, humans responded to horizontal expansions or contractions of the array by shifting their search vertically, preserving angles from landmarks to goal. Pigeons did not adjust their search vertically. Humans trained with a single landmark adjusted search distance when landmark size was changed. Both pigeons and humans use the configuration of a landmark array, but the underlying processes seem to differ.

Averaging temporal duration and spatial position

Ken Cheng, University of Toronto

Marcia L. Spetch, University of Alberta

Paul Miceli, University of Toronto

Pigeons and humans performed on a task in which spatial position and elapsed time redundantly signaled the availability of reward. On each training trial, a landmark moved steadily across the monitor screen. After a fixed amount of time and movement, reward was available for a response. On occasional unrewarded tests, the landmark moved at 0.50, 0.75, 1.00, 1.50, or 2.00 times the training speed. In both pigeons and humans, the central tendency in the response distribution on tests differed across speeds, when measured in terms of both elapsed time, and landmark position. Pigeons and humans seem to average a duration of time and a spatial position to come up with a single criterion time-place that corresponds to the expected time-place of reward.

Procedural determinants of coding processes in pigeons' memory for duration

Marcia L. Spetch, Douglas S. Grant and Ronald Kelly

University of Alberta

Pigeons' retention functions for duration samples differ qualitatively in choice and successive delayed-matching-to-sample tasks. This research tested procedures designed to be hybrids of these tasks. In Experiment 1, adding a fixed-interval component to the test phase of the choice procedure did not eliminate the "respond- short" effect that is characteristic of retention functions for duration in the choice task. A respond-short effect was not present after the birds were subsequently trained in the successive task. In Experiment 2, a choice component was added to the successive task by providing an option stimulus that could be selected to obtain reinforcement on S- trials that followed either short or long samples. Pigeons showed a respond-short effect under this successive-option procedure but did not show a respond-short effect after training with the option stimulus removed from the successive procedure. Thus, the different retention functions obtained in choice versus successive tasks do not appear to reflect differences in the temporal aspect of the test stimulus schedule or the successive versus simultaneous viewing of the test stimuli. Instead, a respond-short effect emerges when subjects make a choice response based on sample duration, but not when they make a go/no-go response based on sample duration.

Overshadowing in Landmark Learning: Touch-Screen Studies with Pigeons and Humans

Marcia L. Spetch

University of Alberta

Overshadowing in landmark learning was investigated in both pigeons and undergraduate students using a touch-screen spatial search task. Ss searched for an unmarked goal presented in varied locations on a computer screen. Graphic stimuli served as landmarks. The effect of the presence of other landmarks on the control acquired by a given landmark was assessed using a design in which each S was trained with 2 sets of landmarks. Both pigeons (Experiment 1) and humans (Experiments 2-4) showed evidence of learning more about a landmark that was the closest landmark of its set to the goal than about a landmark that was of equal distance to the goal but was not the closest landmark of its set. That is, control by a landmark was overshadowed when it occurred together with a landmark that was closer to the goal. Landmark effectiveness appears to depend not only on the absolute properties of a landmark, but on relative factors. The relevance of basic principles of associative learning to spatial landmark learning is discussed.

Stimulus Control in the Use of Landmarks by Pigeons in a Touch-Screen Task

Ken Cheng, University of Toronto

Marcia L. Spetch, University of Alberta

Pigeons were tested in a search task on the surface of a monitor on which their responses were registered by a touch-sensitive device. A graphic landmark array was presented consisting of a square outline (the frame) and a colored "landmark". The unmarked goal, pecks at which produced reward, was located near the center of one edge of the frame, and the landmark was near it. The entire array was displaced without rotation on the monitor from trial to trial. On occasional no-reward tests, the following manipulations were made to the landmark array: (1) either the frame or the landmark was removed; (2) either one edge of the frame or the landmark was shifted; and (3) two landmarks were presented with or without the frame present. On these two-landmark tests, the frame, when present, defined which was the 'correct' landmark. When the frame was absent, the 'correct' landmark was arbitrarily determined. Results showed that pecks of 2 pigeons were controlled almost solely by the landmark, pecks of 3 were controlled primarily by the landmark but the frame could distinguish the correct landmark, and 1 bird's behavior was controlled primarily by the frame. Stimulus control in this search task is thus selective and differs across individuals. Comparisons to other search tasks and to other stimulus control experiments are made.

Pigeons' Use of Landmarks Presented in Digitized Images

Marcia L. Spetch, University of Alberta

Donald M. Wilkie, University of British Columbia

Pigeons received food for pecking a hidden goal location in digitized images of an outdoor scene, presented on a monitor/touchscreen system. Three landmarks (tree, flowers, log) were located near the goal on a field of grass. For some birds the goal was closest to the tree, whereas for other birds the goal was closer to the flower and log. After training, the landmarks were removed, shifted or altered using imaging software. Landmark control was highly selective, and depended on proximity to the goal. There was no evidence that search was controlled by the configuration of the three landmarks. Instead, search behavior appeared to be jointly controlled by a single local landmark and global cues associated with the area of the screen that contained goal locations during training. These findings extend results of a previous study that used simple graphic forms as landmarks in a touch-screen task, and they demonstrate the use of video imaging technology for studies of spatial landmark learning.

The Role of Asymmetrical Coding of Duration Samples in Producing the Choose-Short Effect in Pigeons

Douglas S. Grant and Marcia L. Spetch

University of Alberta

Pigeons were trained on two independent matching-to-duration-samples tasks; one involved 2- and 10-s durations and color choice stimuli, and the other involved 4.5- and 22.5-s durations and line choice stimuli. Accuracy was above chance on mixed-choice probes in which either of the short-duration samples was followed by the two short-associated stimuli. Following explicit training on mixed-choice trials involving choice between the two short- and the two long-associated stimuli, a choose-short effect was demonstrated with both sets of duration samples. These findings are inconsistent with the possibility that the choose-short effect reflects processes of asymmetrical- sample coding and default responding.

Determinants of pigeons' choice between certain and probabilistic outcomes.

Marcia L. Spetch, Michael V. Mondloch, Terry W. Belke

University of Alberta

Roger Dunn

San Diego State University

Pigeons chose between 50% and 100% reinforcement on a discrete-trials concurrent-chains procedure with fixed-ratio 1 initial links and fixed-time terminal links. The 100% alternative always provided food after a terminal-link delay whereas the 50% alternative provided food or blackout equally often after a delay. Additionally, the terminal-link stimuli on the 50% alternative were correlated with the outcomes in signalled, but not in unsignalled, conditions. The effects of intertrial interval duration and length of the terminal-link delays on choice of the 50% alternative were investigated in four experiments. Preference for the 50% alternative varied with signal condition and duration of the terminal link leading to food, but not with intertrial interval or duration of the terminal link leading to a blackout. The results are discussed in terms of conditioned reinforcement effects, Mazur's hyperbolic decay model, and delay reduction.

Mediated Transfer Testing Provides Evidence for Common Coding of Duration and Line Samples in Many-to-One Matching in Pigeons

Douglas S. Grant and Marcia L. Spetch

University of Alberta

A three-phase transfer design was used to determine whether pigeons use a single, common code to represent line and duration samples which are associated with the same comparison stimulus. In Phase 1, two sets of samples (two lines and two durations) were associated with either a single set of comparisons (group MTO, many-to-one) or with different sets of comparisons (group OTO, one-to-one). In Phase 2, one set of samples was associated with a new set of comparisons. In Phase 3 (transfer test), the alternate set of samples was substituted for the Phase 2 samples. Group MTO, but not group OTO, demonstrated immediate transfer. It was concluded that associating a line and a duration sample with the same comparison stimulus results in those samples being represented by a single code.

Control of Pigeons' Spatial Search by Graphic Landmarks in a Touch-Screen Task

Marcia L. Spetch and Michael V. Mondloch

University of Alberta

Pigeons learned to peck an unmarked 2 cm square target area, defined by 4 visually distinct graphic landmarks, on a color monitor with an attached touch frame. The configuration of landmarks and target area was constant during training, but their location on the screen varied across trials. The presence, relative location, and features of the landmarks were manipulated on probe trials. Most birds showed control by only 1 or 2 of the landmarks, and some birds displayed surprisingly accurate search with a single landmark. For individual birds, landmark-removal tests were very consistent with landmark-shift tests in indicating which landmark or landmarks controlled search. However, the dominant landmark varied across birds. Manipulation of landmark color and shape revealed that control was based exclusively on color.

Memory for Duration in Pigeons: Dissociation of Choose-Short and Temporal-Summation Effects

Douglas S. Grant and Marcia L. Spetch

University of Alberta

Five groups of pigeons were trained in a symbolic choice-matching task involving short (2 s) and long (10 s) durations of houselight as samples. Four groups also received training with a second set of samples: line orientations or 2- and 10-s presentations of keylight. The type of sample-to- comparison mapping varied across groups. Although only two of the five groups demonstrated a choose-short effect (a tendency to choose the comparison associated with a short sample at longer delays), all groups demonstrated temporal summation (a tendency to respond on the basis of the combined duration of two successively presented samples). Moreover, the magnitude of temporal summation was equivalent in groups which did and did not demonstrate a choose-short effect. The results suggest that the processes underlying the perception of sample duration remain invariant across different sample-to- comparison mapping arrangements, but that the memory code used to retain temporal information varies.

Pigeons' Memory for Event Duration in Choice and Successive Matching-to-Sample Tasks

Marcia L. Spetch and Douglas S. Grant

University of Alberta

Pigeons were trained to discriminate between short (2 s) and long (10 s) houselight samples within successive or choice delayed matching-to-sample tasks, and then were given four types of tests. During one type of test, the delay interval between the sample and test period was manipulated within sessions. Consistent with earlier findings, pigeons in the choice task displayed a reliable choose-short tendency at longer delays, whereas those in the successive task failed to show an analogous respond-short tendency. A second type of test entailed multiple sample presentations: the short and long samples were preceded by short or long presamples. Pigeons in both the choice and successive tasks showed a temporal-summation effect rather than a sample-consistency effect. In a third type of test, the intertrial interval was manipulated within sessions. Pigeons in both tasks showed a respond-long bias when the intertrial interval preceding the trial was shortened. In a final test, pigeons received no-sample tests, and tests with a 40-s sample. Pigeons in both groups showed a strong tendency to respond short on no-sample tests, and to respond long following a 40-s sample. It appears that timing processes are similar in the two tasks, but that memorial processes differ.

Analogical and Nonanalogical Coding of Samples Differing in Duration in a Choice-Matching Task in Pigeons

Douglas S. Grant and Marcia L. Spetch

University of Alberta

When trained in a symbolic choice-matching task involving short (2-s) and long (10-s) durations as samples, pigeons tend to choose the comparison associated with a short sample as delay increases (choose-short effect). The present experiments showed that the choose-short effect can be eliminated by training in many-to-one (MTO) procedures in which 2 or more sets of sample stimuli are associated with 1 set of comparison stimuli. It is concluded that (a) the choose-short effect results from a process of subjective shortening that occurs during a delay if duration samples are coded analogically and (b) samples of duration are coded nonanalogically in at least some MTO mapping arrangements.