Diversity within Spatial Cognition: Memory Processes Underlying Place Recognition LINDA J . ANOOSHIAN AND PENNIE S. SEIBERT Boise State University, USA SUMMARY We conducted three experiments to explore distinct memory processes involved in remembering places in spatial environments. The results of all three experiments demonstrated the viability of the process-dissociation procedure for studying spatial cognition; that procedure yielded separate measures of the role of familiarity (implicit memory) and conscious recollection (explicit memory) in recognizing Scenes along a previously viewed route of travel. Those measures were not affected by whether the participants viewed videotapes or also physically walked the route of travel. Increasing the delay between encoding and retrieval led to comparable effects for familiarity and conscious recollection (Experiment 1). In contrast, the adverse consequences of dividing attention during encoding were specific to conscious recollection; familiarity estimates were unaffected (Experiments 2 and 3). Overall, the results reinforced the viability of process dissociation as a vehicle for exploring diverse memory processes underlying place recognition. This research was designed to explore diversity within spatial cognition, and to explore different memory processes involved in coming to know large-scale spatial environments. In addressing this diversity, we focused on place recognition, an essential component of successful way finding in large-scale environments (Allen, Siegel and Rosinski, 1978; Chown, Kaplan and Kortenkamp, 1995; Cornell, Heth and Alberts, 1994; Kirasic, Siegel and Allen, 1980). Emphasis on place recognition in theories of spatial cognition has a long history-from Siegel and Whites (1975) seminal theory of the development of spatial cognition to Chown et d . s more recent scene-based model of cognitive mapping and Cornell et al.s (1994) model of way finding by place recognition. In the latter model, people rely on the familiarity associated with particular places when making navigational decisions. For example, to duplicate a route of travel, one need only approach those places that are most familiar. To reverse a route, it works best to approach places with intermediate familiarity. Like Cornell et al.s model, our research reflects the assumption that successful navigation does not always involve conscious or strategic use of organized These experiments were supported by a grant from the Faculty Development Committee at Boise State University to the first author. Special thanks are extended to Edith Meyers for her special help in coordinating and conducting the testing of subjects and to Gene Williams for developing the computer programs. Thanks are also due to J an Fletcher, Gaylen Pack, Craig Prescott and Deborah Thiebert. Requests for reprints should be sent to Linda J. Anooshian, Department of Psychology, Boise State University, Boise, ID 83725, USA. CCC 08884080/96/04028 1-19 0 1996 by J ohn Wiley & Sons, Ltd. Received 2 June 1995 Accepted 10 October 1995 I 282 L. J. Anooshian and P. S. Seibert spatial representations. Specifically, we propose that place recognition can reflect implicit memory or unconscious retrieval as well as explicit memory or conscious retrieval (e.g., J acoby, Yonelinas and J ennings, 1995). The present experiments were designed to disentangle the contributions of different memory processes to recognizing scenes along routes of travel. Specifically, we focus on procedures for obtaining measures of both familiarity (implicit or unconscious retrieval) and conscious recollection (explicit or conscious retrieval) in place recognition. As one of us has argued elsewhere, spatial knowledge has typically been assessed with explicit-memory tasks (Anooshian, 1988; see also Schacter & Nadel, 199 1). With the emphasis on memorability (conscious recollection), landmarks and landmark knowledge have often been defined and/or measured with recognition and/or recall tasks (see Presson, 1987a; Siegel, Kirasic and Kail, 1978). For example, memorability appears to be the primary criterion for determining which of many places in environmental spaces are selected as landmarks in our cognitive maps; landmarks are the places that people recognize, places they recall, or places that they recall first-i.e., places with special status in spatial memory (Presson, 1987a). When researchers ask people to draw maps, the landmarks in those maps are necessarily the places that people consciously recollect when deciding what to include in their maps. Much of the spatial cognition literature has focused on developing more sophisticated methods for assessing spatial knowledge that, unlike maps or verbal descriptions, are not confounded by individual differences unrelated to spatial knowledge (e.g., drawing ability; see Anooshian & Siegel, 1985). Yet, these methods also assume conscious recollection or explicit memory as a minimal criterion for landmark knowledge (see Anooshian and Kromer, 1986). Consistent with Cornell et al.s (1994) emphases on familiarity and place recognition in way finding, we endorse greater attention to nonstrategic or implicit uses of memory in studies of spatial cognition (see also Chown et a[., 1995). For the present experiments, estimates of familiarity were obtained to reflect implicit or unaware memory retrieval processes; estimates of conscious recollection reflected explicit or conscious memory-retrieval processes (as in J acoby, 1991; J acoby and Kelley, 1991). Despite the focus on conscious recollection or explicit memory in the literature on spatial cognition, everyday experiences emphasize the importance of familiarity or implicit remembering (Anooshian, 1988; Cornell et al., 1994). For example, when finding our way along once-familiar routes, we usually remain confident as long as the surroundings seem familiar. Without such a feeling of familiarity, we are apt to conclude that we have taken a wrong turn (see also Cornell et al., 1994). Yet, had we been asked to recall landmarks or to estimate distances between places (e.g., as in Anooshian and Kromer, 1986; see Siegel, 1981), specific landmarks just would not have come to mind. Our research was designed to establish a methodology for disentangling the contributions of explicit- and implicit-memory processes to place recognition. It seemed that such a methodology should have a number of significant pay-offs for spatial-cognition researchers. First, as suggested earlier, a better understanding of the diversity of processes underlying place recognition is likely to yield a better understanding of diversity within everyday navigational experiences. Second, this understanding appears critical in addressing apparent dissociations or independence among different types of spatial knowledge about the same environmental areas. For Diversity within Spatial Cognition 283 example, Anooshian and Kromer (1986) reported that the landmarks to which children made accurate bearing estimates tended to be different than the landmarks about which they could accurately estimate distance. We often know specific salient landmarks, but not necessarily the direction of them or the specific route to them. Dissociations between implicit and explicit memory reflect one possible approach to accounting for the apparent independence of different types of knowledge about spatial areas. That is, spatial memories that rely primarily on implicit memory should appear independent of spatial memories that rely primarily on explicit memory (Schacter and Nadel, 1991). Dissociations between measures of implicit and explicit memory have been well documented with various implicit-memory tasks with normal adults as well as amnesiacs (see reviews by Richardson-Klavehn and Bjork, 1988; Schacter, 1987). For example, in the case of amnesia, it is common to find high levels of implicit remembering, evidenced by improved performance after prior exposure, associated with the inability to remember anything about the prior exposure-even that it occurred. Distinct memory processes may also be useful in addressing diversity in acquisition strategies, and differences in resulting patterns of spatial memory. For example, Presson (1 987b) distinguishes between primary and secondary uses of spatial information. Secondary spatial memory guides symbolic uses of space while primary spatial memory guides practical uses of space such as successful navigation. Direct experiences with environments appear to promote primary memory while exposure to maps promotes secondary memory (e.g., as reflected in directional or Euclidian distance estimates, effects of orientation manipulations; Evans and Pezdek, 1980; Presson, DeLange and Hazelrigg, 1989; Thorndyke and Hayes-Roth, 1982). Such findings have been taken as evidence that different memory mechanisms underlie primary and secondary spatial memory (e.g., Schacter and Nadel, 1991). It certainly seems possible that different ways of representing spaces, as associated with different acquisition approaches or strategies, are differentially reliant on implicit- versus explicit-memory processes. Rather than focusing on dissociations between memory measures derived from different tasks, we attempted to develop procedures that would provide separate measures of implicit and explicit processes within the same task. The specific procedure for doing so, process dissociation (J acoby, 1991), represents a major departure from the typical research approach used to examine implicit and explicit memory-the approach of administering different tasks to obtain measures of different memory processes (see kchardson-Klavehn and Bjork, 1988). J acoby (1991) argues that no single task can provide a pure measure of memory process; most task performances reflect some combination of implicit and explicit processes. For example, when identifying a word as old in a recognition task (usually considered an explicit memory task), participants may sometimes rely on a general feeling of familiarity typically associated with implicit remembering. Of course, it is usually difficult to differentiate between memory processes with a single task since, for most tasks, both processes facilitate the same performance or yield the same responses. For example, if asked to identify words or places seen before, either conscious recollection of the prior episode (explicit memory) or a sense of familiarity (implicit memory) would yield the same conclusion (Ive seen it before.). Similarly, in navigating through a large-scale space, conscious recollection of an upcoming landmark has the same consequence as a feeling of familiarity; in either case, one is 284 L. J. Anooshian and P. S. Seiberi likely to conclude that one is headed in the right direction. The process dissociation procedure, developed by J acoby for word recognition, involves comparing levels of performance when familiarity and conscious recollection yield the same conclusions and when the effects of the two memory processes act in opposition, yielding different conclusions. In our place-recognition task, participants chose between video frames viewed in earlier exposure to a route of travel and similar video frames that had been changed through computer editing. Our interest was in the proportion of route frames that participants selected when instructed to select such frames (inclusion instructions) and when instructed not to select such frames (exclusion instructions; putting the two memory processes in opposition). Then, estimates for conscious recollection and familiarity were based upon the following assumptions about the selection of route frames in the inclusion (P[INC]) and exclusion conditions (P[EXC]): P[INC] = P[CONS] + P[FAMIL] - (P[CONS] x P[FAMIL]) The proportion of route scenes selected in the inclusion condition should reflect the combined probabilities of conscious recollection (P[CONS]) and sense of familiarity (P[FAMIL]), adjusting for the probability of joint occurrences (already counted in both other probabilities); and The proportion of route scenes selected in the exclusion condition should reflect the probability of the joint occurrence of familiarity and the absence of conscious recollection. Task instructions for exclusion emphasized this joint occurrence (e.g., choose a familiar-looking scene as long as you do not specifically remember it from the original route). Without providing the details of the algebraic solutions, it should be clear that one can derive formulae for the probability of familiarity and of conscious recollection from the algebraic assumptions listed above. As outlined by J acoby (1991), performances in the two conditions (inclusion and exclusion) are used to derive estimates for conscious recollection (P[CONS]) and familiarity (P[FAM]): P[EXC] = P[FAMIL] x (1 - P[CONS]) P[CONS] = P[INC] - P[EXC] P[FAMIL] = P[EXC]/(l -P[INC] + P[EXC]) A major purpose of our research efforts has been to demonstrate the viability of a method like process dissociation to the study of spatial task performances. Although some have questioned the assumption of independence upon which process dissociation relies, past reviews of literature on implicit and explicit memory provide ample and compelling evidence of memory processes that are independent and dissociated (J acoby et al., 1995; Richardson-Klavehn and Bjork, 1988; Schacter, 1987). In addition to providing measures that reflect diversity within a single task performance, process dissociation is designed to explore functional dissociations between different memory processes. Such functional or process dissociations are demonstrated when specific experimental manipulations have different effects on measures of familiarity and conscious recollection. For Experiment 1, we manipulated both the nature of initial exposure to the route of travel and the delay between exposure and test (with inclusion and exclusion conditions). For Experiments 2 and 3, we examined the effects of dividing attention on measures of familiarity and conscious recollection obtained after exposure to different types of environmental areas. Diversity within Spatial Cognition 285 EXPERIMENT 1 In designing Experiment 1, wepredicted that familiarity estimates, like measures of primary uses of spatial information (Presson, 1987b), would be more dependent on direct experiences with the physical environment than would be the case for conscious recollection. In varying the nature of the initial exposure to the route of travel in Experiment 1, some participants actually walked the route as well as viewed it on videotape; others simply viewed the video presentation. We also predicted that conscious recollection, but not familiarity, would be adversely affected by increasing delay between exposure and test. Past research has demonstrated that performances on implicit memory tasks are sometimes considerably more resistant to decay than explicit-memory measures (e.g., recognition or recall; Mitchell and Brown, 1988; Musen and Treisman, 1990; Sloman, Hayman, Ohta, Law and Tulving, 1988). Method Participants Participants were volunteers from a General Psychology participant pool of a public university in the northwest. A total of 12 females was tested in each of 3 delay conditions (immediate, 2 days, 1 week), with half randomly assigned to each exposure condition (walk-view versus view-only). We had intended to include both males and females but failed to get a sufficient number of male volunteers for delay conditions requiring two testing sessions. We excluded participants who had worked or taken a class in the technology building from which our experimental materials were obtained. Overview of procedure The three phases of the experimental procedure reflected the essential elements of process dissociation (J acoby, 1991). In Table 1, we have summarized the general correspondence between the sequence followed by J acoby (1991) and in this experiment. As reflected in that table and summarized in later sections, there were a number of changes or differences required to adopt the J acoby procedure, as developed for word recognition, to place recognition. Video materials Most task materials were derived from a videotape of a route of travel (Route 1) through two floors of a technology building infrequently visited by university students. For the original filming, the camera-person sat in a wheelchair that was moved at average walking speed along the route; the complete video lasted 6 minutes. A total of 80 frames spaced approximately evenly along the route was captured from this videotape and stored as picture files (on a computer hard drive, using SVIA software). Half of these 80 frames were randomly selected and designated for computer-editing to make new items for the third test phase (see Table 4). This computer editing (using Photostyler software) involved between two and four minor changes for each frame (e.g., adding/enlarging door, adding/ removing window), thereby insuring that the overall configuration of the original videotaped scene remained relatively intact. We used edited video frames in lieu of new frames for two related reasons. First, the video of the route of travel contained 286 L. J. Anooshian and P. S. Seibert Table 1. Phases of the experimental procedure from J acoby (1991) as adopted for Experiment 1 Jacoby (1991Fword Phaseltest recognition Experiment 1-place recognition Phase I Phase 2 Phase 3- Recognition test Inclusion instructions Exclusion instructions READ. . . Read words LISTEN . . . Listen to second set of words, to prepare for later recognition test (Phase 1) versus new words ROUTE 1 . . . View video and/or slide show of ROUTE 2 . . . View scenes that presumably Route 1 contain objects from another route, Route 2, to prepare for later test Critical trials =read words Critical trials =Route 1 frames (Phase 1) versus edited Route 1 frames (forced choice) Select any word in prior Select any scene from Route 1 or any scene lists . . . select read words containing a Route 2 Object . . . select Route 1 scenes Object . . . do not select Route 1 scenes. If remember scene as exactly like scene from Route 1, it is definitely new or incorrect Select only words that had Select only scenes that contain a Route 2 been heard . . . do not select read words. If remember as word that was read, it is def ~t el y new choice near-complete exposure to hallways in the selected buildings. Second, we anticipated that pairing old scenes with scenes from a different building, even if similar to the original Phase 1 scenes, would yield ceiling levels of performance often associated with picture recognition. The video and video frames described thus far were used for exposure to Route 1 (Phase 1) and for the critical trials of the third test phase (see Table 1). Additional video segments were taken at diverse locations around the same university campus for Phase 2. Locations for these segments were selected on the basis of ( 1) general similarity to Route 1 scenes (e.g., age of fixtures, general architectural style), and (2) the existence of at least three objects (e.g., door, window, cabinet). Of the 70 frames captured from these video segments, 50 were randomly selected for use as Route 2 Object scenes for the second exposure phase (Phase 2). Each of these 50 frames presumably contained at least one object that could be seen somewhere along Route 2. The remaining 20 frames were used as new items in recognition testing. Finally, 20 of the 50 Route 2 Object frames were randomly selected for inclusion in the third test phase as old items. As detailed above, the various frames captured from videotapes included 120 frames used in recognition testing: 40 Route 1 frames, 40 edited Route 1 frames, 20 Route 2 Object frames, and 20 new frames. Some additional video frames were captured (and some edited) for use as examples to supplement task instructions. Procedure Individually tested, each participant was exposed to a route of travel in one of two exposure conditions, walk-view or view-only. The last two phases of the procedure, separated by one of three delay intervals (lOmin, 2 days, 1 week), allowed for the Diversity within Spatial Cognition 281 application of Jacobys process-dissociation procedure (J acoby, 199 1) to place recognition. Phase I : exposure to Route 1. Participants in the walk-view exposure condition first followed the experimenter along the entire course of the route of travel with instructions to pay attention so that they could later find their way along the route without help from the experimenter. Upon returning to the testing room, the participant viewed the videotape of the route twice. Finally, the participant led the experimenter along the actual route of travel; if necessary, the experimenter corrected wrong turns. For the view-only condition, video presentations were substituted for actual walks along the route; participants viewed the videotaped route four times. During the first viewing, the participant traced a finger along a map of the area that marked the route with arrows. During the final viewing, the participant traced a finger along a map of the area without the marked route, presumably as a test of their learning. As for the walk-view condition, the experimenter corrected wrong turns. Phase 2: exposure to 50 other scenes-Route 2 Object scenes. With the use of two example frames, participants were told that they would see many potential Route 2 Objects-that is, doors, windows, cabinets, etc. identical to ones seen somewhere along the hypothetical Route 2. Specifically, they were told that at least one such Route 2 Object could be seen in each of the 50 frames that they would view. The task was explained as a test of a new theory that simple exposure to objects such as windows or doors along a route of travel, even if intermixed with lots of scenes and objects not found along the route, would facilitate later route learning. Task instructions emphasized that the 50 Route 2 Object frames were not scenes from Route 2; participants were told that these particular frames or scenes had been selected because they contained at least one object seen somewhere along Route 2. Each participant saw 50 video frames for 2sec each, with periodic reminders that only one object in a single frame was an object from Route 2, Memory for frames from this second exposure phase was not considered when deriving memory estimates with process dissociation. Rather, memory estimates were derived entirely from the proportion of Route 1 frames selected over edited Route 1 frames under different instruction conditions. Under what conditions would participants select a Route 1 frame when told not to do so-i.e., under exclusion instructions? Our second exposure phase was designed with this question in mind. Our intent was to expose participants to a wide array of potential objects; in later testing, they would be given the difficult task of identifying scenes that contained some object (e.g., window) from our hypothetical Route 2. We designed Phase 2 such that, from the participants perspective, almost any video frame seen in a later testing session (Phase 3Fparticularly if it seemed familiar--could conceivably contain at least one Route 2 Object. Our intent was to avoid arbitrary response patterns that could otherwise occur if participants were confident about rejecting both recognition options (e.g., under exclusion instructions, identifying one as a Route 1 frame but the other as also incorrect). Phase 3: testing using process dissociation. Each participant completed both the inclusion and exclusion tasks. For 20 critical trials in each task, the participant chose 288 L. J. Anooshian and P. S. Seibert between a Route 1 frame and an edited Route 1 frame. For 10 additional trials in each task, the options for forced-choice recognition included a Route 2 Object frame from the second exposure phase and a new frame. Forced-choice recognition was used because pilot testing had revealed extreme patterns of responding when only one picture was presented at a time; many participants resorted to either accepting or rejecting all frames presented to them. For each participant, half of each type of frame (20 Route 1,20 edited Route 1, 10 Route 2 Object, 10 new frames) was randomly assigned to each of the two tasks (inclusion, exclusion). Within each task, the specific pairing of frames for recognition trials (Route 1 with edited Route 1; Route 2 Object scene with new frame), the position of each frame on the computer screen (top or bottom), and the order of the recognition trials were also randomly determined for each participant. Pilot testing had revealed some difficulties in providing clear instructions for the exclusion task when administered second, following inclusion. It appeared that participants had difficulty rejecting Route 1 scenes after completing a task in which they had been designated as correct choices. Hence, for this experiment, the exclusion task was always administered first. The exclusion task was the Route 2 Objects task. Consistent with the aims of process dissociation, this task was designed to estimate how frequently participants would select Route 1 frames when explicitly instructed not to do so. Specifically, this Route 2 Objects task was designed so that Route 1 selections reflected the presence of familiarity and the absence of conscious recollection. Participants were told to select scenes that contained at least one object that could be seen somewhere along Route 2 (i.e., seen in one of the Route 2 Object scenes). Given the obvious difficulty of this task, instructions emphasized that there were only two types of frames about which participants could be certain: (1) unedited Route 2 Object frames (to be selected), and (2) unedited Route 1 frames (to be excluded or rejected). Participants were told again, as in prior instructions, that Route 1 and Route 2 had no specific objects in common. For other frame possibilities, the experimenter advised participants to select the one scene that looked most familiar since it probably contained a Route 2 Object. Task instructions contained repeated reminders that participants were to select familiar-looking scenes as long as they were not unedited frames from Route 1. Our inclusion task was the complete scenes task. Participants were told to select scenes/frames that they had seen before in either of the two exposure phases. Examples were used to warn participants about the existence and general nature of the computer editing of video frames. Participants were told to select only unedited frames exactly like those seen before. Results and Discussion Preliminary analyses revealed no reliable effects involving exposure condition (walk- view versus view-only). The only effect that approached significance was the effect of exposure condition on inclusion scores; participants in the walk-view exposure condition performed more poorly, selecting fewer old route scenes ( M = 0.64) than did those in view-only condition (M=0.71; P=O. lO). We mention this only to emphasize that there were no adverse consequences for not being able to experience the physical environment directly. On the one hand, these results were somewhat Diversity within Spatial Cognition 289 disappointing in failing to demonstrate a functional dissociation between familiarity and conscious recollection. On the other hand, the results served an important role in directing further research. For Experiments 2 and 3, we could be confident that video presentations served as reasonable simulations of direct experiences with routes of travel (at least for place recognition). In fact, our results are consistent with other research indicating that direct experiences are not necessarily more helpful than simulations in acquiring spatial knowledge (Hunt, 1984; Kirasic and Mathes, 1990). For further analyses, we used ANOVA designs with a single between-subject factor for delay between exposure and test (lOmin, 2 days, 1 week). Dependent measures for memory tasks included an additional within-subject factor for task (inclusion, exclusion) or derived memory estimate (recollection, familiarity). The significance level was set at p < 0.05. Memory measures Initial dependent measures for the memory tasks reflected the proportion of Route 1 frames selected in the two tasks: inclusion (when participants were told to select route scenes) and exclusion (when participants were told not to select these route scenes). Consistent with the aims of process dissociation, these proportions were used to derive separate estimates for the role of conscious recollection and sense of familiarity in identifying route scenes. These two estimates were calculated, individually for each participant, using the following derived formulae: P[CONS] = P[INC] - P[EXC] P[FAMIL] =P[EXC]/(l -P[INC] + P[EXC]) These derived estimates were analysed as a within-subject factor in an ANOVA design. That analysis revealed only a reliable effect for memory estimate, F(1,33) = 98.80, MSe = 0.02; estimates for familiarity (M=0.57) were higher than estimates for conscious recollection (M= 0.24). As can be seen in Figure 1, there was a relatively steady decline in both memory estimates across increasing delay. But the main effect for delay was only marginally significant (p = 0.06). Overall, the results of Experiment 1 failed to provide evidence of functional dissociations between measures of familiarity and conscious recollection. Richardson-Klavehn and Bjork (1988) pointed out that inconsistent results have been associated with research on delay effects using explicit and implicit memory tasks. In discussing such inconsistencies, Musen and Treisman (1 990) suggested that persistence across delay for implicit measures (e.g., priming) is specific to visual stimuli with uniqueness, novelty and specificity of visual details. Hence, persistence across delays for familiarity measures may be more apparent for nonsense stick figures, as used by Musen and Treisman, than for scenes from everyday environments, as used in this experiment. Route 2 selections Thus far, the analyses of memory scores have been based on performances for the 20 critical trials within each task; that is, measures were based on the proportion of trials for which the participants selected old Route 1 frames when paired with edited Route 1 frames. The 10 additional trials within each task provided measures of the proportion of trials for which participants selected a frame from the second exposure phase when paired with a new frame. Of course, these Route 2 Object frames were 290 L. J. Anooshian and P. S. Seiberl Figure 1. Mean memory estimates, for conscious recollection and familiarity, for the different delay groups of Experiment 1. correct choices for both the inclusion and exclusion tasks. Analyses of scores for these additional trials allowed us to evaluate a potential risk or criticism associated with process dissociation. That is, the procedure is based on the assumption that conscious recollection and familiarity are invariant across the inclusion and exclusion tasks. For example, one assumes that participants use the same criterion for including Route 1 frames in inclusion as they do for excluding the same frames in exclusion; in each case, the selection or rejection should reflect successful conscious recollection (J acoby, 1991). An ANOVA on the proportion of trials for which Route 2 Object frames were selected revealed no reliable effects; the proportion of old frames selected was comparable for the inclusion task ( M= 0.82) and the exclusion task ( M= 0.86). This result provides suggestive evidence that memory processes were relatively invariant across the exclusion and inclusion tasks; it is unlikely that participants were more cautious in accepting old frames in the exclusion compared to the inclusion task. Although these data are admittedly suggestive rather than definitive, it is encouraging and relevant that there have been numerous demonstrations of the correspondence between the empirical findings obtained from process dissociation and other methodologies (see J acoby et al., 1995; LeCompte, 1995). Diversity within Spatial Cognition 29 1 The overall results of this experiment, although providing no evidence of functional dissociations between familiarity and conscious recollection, were nevertheless promising in pointing to the viability of process dissociation as a method for studying diverse memory processes reflected in the completion of a single spatial task. This viability was further reinforced by the results of our next two experiments. EXPERIMENT 2 For this experiment, we attempted to demonstrate functional dissociation between familiarity and conscious recollection by manipulating attention during initial exposure. J acoby and Kelley (1991) successfully used process dissociation with word recognition to demonstrate that the effects of dividing attention were specific to conscious recollection; familiarity estimates were not affected. If this finding generalized to scene recognition, there would be important implications for place memory and way finding. If distracted and not attending to a known route of travel, individuals may still retain a sense of familiarity associated with their surroundings. Familiarity is often an essential component of successful way finding (Cornell et al., 1 994). We suggested earlier that diversity in memory processes was relevant in accounting for diversity in strategies for acquiring spatial knowledge (e.g., Thorndyke and Hayes-Roth, 1982). In this experiment, we explored relations between memory estimates derived from process dissociation and individual differences in self-reports of spatial strategies. Theories of cognitive mapping typically emphasize how the encoding of landmarks ultimately leads to configurational or survey representations of environmental areas. In fact, Kuipers (1982) described these theories as assuming a map in the head metaphor. At the end of the testing session for Experiment 2, we asked participants to indicate the extent to which they had tried to learn the original route in terms of a configuration or map in the head. This acquisition strategy is assumed to depend on effortful processing and to be demanding of attention (Anooshian and Siegel, 1985). Hence, weexpected that reports of learning routes in terms of configurations would be more closely related to conscious recollection than familiarity. Method Participants A total of 24 participants was selected from the same General Psychology participant pool used for Experiment 1. Half of the participants (6 males and 6 females) were assigned to each attention condition; task order was randomly assigned. Procedure The procedure was the same as described for Experiment 1 with four major exceptions. Two of those exceptions are described in later sections on the attention manipulation and strategy ratings. Thirdly, weused a streamlined exposure sequence to reduce the boredom that seemed to be generated by the long exposure sequence for Experiment 1. Rather than four viewings of the videotape or two viewings and two physical walks, initial exposure to the route consisted of two viewings of the 292 L. J. Anooshian and P. S. Seibert videotaped route of travel separated by exposure to 80 consecutive graphic images (captured frames) of the route shown on the computer monitor. This slide show of the route of travel included the 40 route frames that would be included as old frames in later recognition trials. Each frame appeared on the monitor for 4sec until replaced by the next frame. Finally, we conducted extensive pilot testing to work out procedures that would allow for varying the order of administering the inclusion and exclusion tasks. To minimize confusion associated with completing the exclusion task last, as observed in earlier testing, we added a number of specific example frames that were integrated with instructions. For Experiment 2, participants were assigned to one of two task orders for the final test phase; inclusion testing was followed by exclusion, or vice-versa. Attention manipulation. Participants in the full-attention condition were tested in the same fashion as described for Experiment 1. The focus was on learning the route of travel (e.g., as diagrammed in maps of the spatial area). For the dwided-attention group, the participants attended to a visual-imagery task, the primary task, while incidentally monitoring turns in the route of travel. For the imagery task, the experimenter briefly flashed a block letter in front of the participant, with dotted lines indicating an imaginary walk around the outside of the letter; block letters were used to increase the number of turns around corners. The participants task was to call out the directions for turning (e.g., right-left-left-right, etc.) if actually walking the routes around imaginary letters. If the participant erred, the experimenter quickly corrected the error and asked the participant to continue. Participants continued calling out directions of turn in their travel around letters throughout the video and slide-show presentations of Route 1; participants completed an average of 12.42 letters. Two aspects of the procedure were designed to ensure that participants, although focusing attention on the primary task, maintained visual fixation on Route 1. First, the experimenter showed block letters for the imagery task very quickly and straight-ahead so as to distract minimally from visual focus on the route of travel. Second, the participants secondary task was to press the button on a computer mouse whenever they observed a change in camera direction in the route exposure. Except for this attention manipulation in Phase 1, procedures for remaining phases were the same for both attention conditions. Strategy ratings. At the end of the experimental session, participants rated their level of agreement with statements about using configurations to learn the original Route 1 (from 1, strongly disagree, to 5, strongly agree). Two statements described trying to form a map in my head to keep track of directions and locations, and forming visual images. Results and Discussion Preliminary analyses revealed no reliable effects for task order or for gender; males and females in the two order groups were combined for further analyses. All dependent measures were analysed with ANOVA designs with a between-subject factor for attention (full or divided) and a within-subject factor for task (inclusion, exclusion) or derived memory estimate (recollection, familiarity). The significance level was set at p < 0.05. Diversity within Spatial Cognition 293 Initial dependent measures reflected the proportion of Route 1 frames selected in the inclusion and exclusion tasks. Analysis of these measures yielded a reliable main effect for task, F( 1,22) = 36.14, MSe = 0.01, and a reliable interaction between task and attention condition, F( 1,22) = 3 1.72, MSe = 0.01. The dividing of attention led tofewer Route 1 selections for inclusion instructions (M=0.55 compared to 0.73 for full attention) and more Route 1 selections for exclusion instructions (M=0.54 compared to 0.34 for full attention); the effect for attention condition was reliable for both inclusion, F( 1,22) = 16.41, MSe = 0.01, and exclusion scores, F(1,22)=15.61, MSe=O.Ol. As for Experiment 1, weused these proportions to derive separate estimates for conscious recollection and familiarity. The analysis of these memory estimates revealed reliable effects of attention condition, F(1,22) = 24.20, MSe = 0.02, and memory estimate, F( 1,22) = 69.35, MSe = 0.02; these effects were interpreted in the context of a reliable interaction between attention condition and memory estimate, F( 1,22) = 20.98, MSe = 0.02. As predicted, the dividing of attention had an adverse effect on conscious recollection (M= 0.02 compared to 0.38 for full attention), F( 1,22) = 3 1.72, MSe = 0.03, but no reliable effect on familiarity estimates (Ms = 0.54 and 0.55 for full and divided attention, respectively). An ANOVA on the proportion of trials for which Route 2 Object frames were selected revealed no reliable effects; the proportion of old frames selected for the inclusion task (M= 0.87) was comparable to that selected for exclusion (M= 0.83). Hence, as also observed for Experiment 1, it appeared reasonable to assume that participants used the same criteria for including Route 1 frames in inclusion as they did for rejecting the same frames in exclusion. Such findings, together with the functional dissociation demonstrated for the manipulation of attention, reinforce the viability of process dissociation as an important approach to exploring the diversity of memory processes underlying spatial cognition. Two self-report ratings of attention to configurations or images in learning the original route were obtained. Correlations with memory estimates were reliable only for the first statement describing forming a map in the head to keep track of directions and distances. For that statement, the correlation with conscious recollection was reliable for the entire group, r (22) = 0.54, and for the full-attention group, r (10) = 0.47. The same correlation was not reliable (r= 0.16) when calculated for the divided-attention group. This pattern of correlations suggests that the dividing of attention may have interefered with the use of effortful strategies for learning routes of travel. Consistent with this suggestion, average ratings for this statement were reliably lower for the divided-attention group (M= 2.58) than for the full-attention group (M= 4.00), F(1,22) = 6.93, MSe = 1.66. These results, although exploratory in nature, are important in suggesting relations between individual differences in estimates for conscious recollection and specific approaches to learning spatial information. In our next experiment, we obtained additional ratings from participants about the approaches that they had used to learn the original route of travel. EXPERIMENT 3 Overall, the combined results of Experiments 1 and 2 were quite promising in terms of reinforcing the viability of process dissociation as an approach for studying the 294 L. f. Anooshian and P. S. Seibert diversity of basic memory processes underlying specific spatial tasks (e.g., place recognition). Given those promising results, it seemed important to obtain additional evidence of their generality. Questions could certainly be raised regarding the extent to which our findings were specific to the type of environment from which our scenes had been taken, the campus technology building. Were our results specific or somehow idiosyncratic to this type of space? To address this question, weselected a very different kind of environment area for Experiment 3, a route through two floors of a major metropolitan hospital. Method Participants A total of 24 participants was selected from the same pool as described for Experiments 1 and 2. Half of the participants (6 males and 6 females) were assigned to each attention condition. Task order, for the inclusion and exclusion tasks, was randomly assigned. Procedure The procedure was the same as described for Experiment 2 with three major exceptions. First, the video and video frames for exposure and testing were obtained from a route through two floors of a metropolitan hospital rather than the campus technology building. Second, wefurther streamlined the first exposure phase (Phase 1). Participants saw three slide shows of 80 captured frames rather than two video viewings separated by one slide show. Third, the statements that participants rated at the end of the experimental session were expanded. Three statements were designed to tap the use of map-like configurations: the two statements described for Experiment 2 and an additional statement indicating that the task of learning Route 1 would have been easier with prior study of maps. Three additional statements focusing on the use of verbal descriptions included (1) making up and remembering verbal directions for travel, (2) making up verbal descriptions of objects and areas along the travelled route, and (3) the task would have been easier with prior study of written directions for travel. These two categories of statements corresponded to two very different, yet equally effective, means for acquiring spatial information (see Kirasic and Mathes, 1990). Results and Discussion Preliminary analyses revealed no reliable effects for task order; the two order groups were combined for further analyses. All dependent measures were analysed with ANOVA designs with between-subject factors for attention (full or divided) and gender as well as a within-subject factor for task (inclusion, exclusion) or derived memory estimate (recollection, familiarity). The significance level was set at p < 0.05. Analysis of the proportion of Route 1 frame selections yielded reliable main effects for gender, F(1,20)=5.33, MSe=O.Ol, and task, F(1,20)=22.36, MSe=O.Ol. These main effects were interpreted in the context of reliable interactions between task and attention condition, F(1,22) = 19.89, MSe = 0.01, and task and gender, F(1,20) = 4.46, MSe = 0.01. Consistent with results for Experiment 2, dividing attention led to fewer Route 1 selections for inclusion instructions (M=0. 56 Diversity within Spatial Cognition 295 compared to 0.72 for full attention) and more Route 1 selections for exclusion instructions (M=0.55 compared to 0.44 for full attention). The attention effect was reliable for both inclusion, F( 1,22) = 9.10, MSe = 0.02, and exclusion scores, F( 1,22) = 8.77, MSe = 0.01. In contrast, the effect for gender was specific to inclusion scores, F( 1,22) = 4.63, MSe = 0.02; females selected more old Route 1 frames (M= 0.70) than did males (M= 0.57). The analysis of derived memory estimates (familiarity and conscious recollection) revealed a reliable main effect for gender, F( 1,20) = 8.45, MSe = 0.02; higher estimates were obtained for females (M= 0.42) than for males (M=0.3 1). Further main effects for attention condition, F( 1,20) = 18.3 1 , MSe = 0.02, and memory estimate, F( 1,20) = 186.07, MSe = 0.01 , were interpreted in the context of the predicted interaction between attention condition and memory estimate, F(1,20) = 11.83, MSe =0.01. Strikingly consistent with the results obtained for Experiment 2, the adverse consequences of dividing attention were specific to conscious recollection (M= 0.01 compared to 0.28 for full attention), F(1,22) = 8.77, MSe = 0.07. For the results for both experiments, the interaction between attention condition and memory estimate reflected floor levels of conscious recollection when attention was divided. Hence, when our attention has been directed elsewhere-a common experience in everyday encounters with large-scale environments-we are likely to retain little explicit awareness of what we have seen. Yet, such adverse consequences of dividing attention do not necessarily prohibit good place recognition. Also consistent with the results of Experiment 2, comparable means were obtained for familiarity (Ms=0.61 and 0.56 for full and divided attention, respectively). The consistency of findings across experiments bolsters our confidence that familiarity is fairly resistant to adverse consequences of divided attention and, hence, is potentially an important component of everyday way finding (see Cornell et al., 1994). Of course, some may argue that the environments chosen for Experiments 2 and 3 were more similar than different, both being indoor and institutional. Hence, despite the consistency of findings, definitive statements regarding the generality of our findings must await further research. An ANOVA on the proportion of trials for which Route 2 Object frames were selected, for 10 trials for each task, revealed no reliable effects; the proportion of old frames selected for the inclusion task (M= 0.78) was comparable to that selected for exclusion (M= 0.75). Hence, as for the first two experiments, there was no reason to suspect that participants used memory processes differently (e.g., being more or less cautious) in the inclusion and exclusion tasks. Earlier results from Experiment 2 had revealed an association between self-reports of relying on configurations in learning Route 1 and conscious recollection estimates. With additional findings from Experiment 3, wewere able to rule out the possibility that these earlier findings reflected associations between conscious recollection and the use of any systematic approach to route learning, whether based on configurations or not. For the additional ratings obtained in this experiment, we averaged ratings for different questionnaire items to obtain two composite scores: agreement with statements about relying on configurations or map-like images, and agreement with statements about relying on verbal descriptions/listings. Consistent with the results of Experiment 2, higher configuration scores were associated with higher estimates for conscious recollection, r (22) = 0.44. Also, configuration ratings were reliably higher for the full-attention (M= 4.11) than for the divided-attention 296 L. J. Anooshian and P. S. Seibert group (M= 3.00), I;( 1,22) = 1 1.72, MSe = 0.54. (No reliable correlations or effects were observed for verbal-strategy scores.) Of course, a variety of explanations are possible in accounting for this pattern of correlational data. Nevertheless, the results are encouraging in suggesting systematic relations between the memory estimates derived from process dissociation and participants own self-reports of their approaches to learning routes of travel. Given that the slide-show exposure for Experiment 3 may implicate visual memory more than spatial cognition, these suggestive results are important in confirming place recognition as an important component of spatial cognition (Chown et al., 1995; Cornell el al., 1994). The gender differences observed in this study-in the proportion of old Route 1 scenes selected for inclusion and in overall memory estimates-were not expected. However, in on-going similar research with cartoon stories rather than routes of travel, wehave obtained evidence of higher familiarity estimates for females than for males, with no differences for conscious recollection (see Anooshian and Seibert, 1996). A full explanation of these gender differences will, no doubt, need to await further experimentation. GENERAL DISCUSSION We contend that an adequate understanding of spatial cognition depends on the development of methodologies designed to assess the combined influences of explicit (conscious recollection) and implicit memory (sense of familiarity) on spatial cognition. We developed such a methodology from research on word recognition, process dissociation (J acoby, 1991; J acoby and Kellery, 1991), and demonstrated its viability in three experiments on place recognition. Using this method, wefound that familiarity estimates were unaffected by the dividing of attention, a manipulation that had fairly dramatic effects on conscious recollection. We suggest that, in a variety of everyday situations, familiarity can guide navigational efforts even when there is little or no conscious recollection of landmarks (Cornell et al., 1994). Overall, our research findings are most important in establishing the nonunity of spatial cognition (Schacter and Nadel, 1991) and in providing a new framework for exploring diversity within spatial memory. The significance of this recommended approach is most apparent in the context of past research on spatial cognition- research that has largely ignored the diversity of memory processes implicated by the emerging literature on implicit and explicit memory. As noted by Schacter and Nadel (1991), standard tests of spatial memory can be characterized by explicit tests: they require intentional, deliberate retrieval of spatial information that was acquired during some prior episodes (p. 180). For the most part, the methods and theories on spatial cognition reflect the assumption that successful navigation relies on conscious recollection of spatial representations (Anooshian, 1988). Process dissociation also offers an almost unlimited range of possibilities for future research. In our experiments, we included manipulations of the nature of exposure, of delay between exposure and test, and of attention on conscious recollection and sense of familiarity. Evidence of process dissociation, or different effects on conscious recollection and familiarity, was obtained only for the manipulation of attention. The manipulations included in our research are but just a few examples of the kinds of manipulations that might reveal interesting Diversity within Spatial Cognition 291 dissociations between different memory measures in future research. As summarized by J acoby et al. (1995), the strength of the process-dissociation procedure lies in the consistency of empirical findings associated with diverse manipulations of cognitive control or conscious intention. For example, beyond the dividing of attention, both aging and retrieval time-variables identified with cognitive control-are associated with variation in conscious recollection, but not familiarity. Together with an extensive research literature documenting dissociations and independence between implicit- and explicit-memory measures, such evidence supports the somewhat controversial assumption of process dissociation that implicit and explicit processes are independent (see J acoby et al., 1995). As also noted by J acoby et al., other possible assumptions about the relation between memory processes (e.g., redundancy or exclusivity) yield results that contradict other empirical findings. In contrast, process dissociation, based on an assumption of independence, yields findings quite consistent with those obtained with other methodologies (Jacoby et al., 1995; LeCompte, 1995). Hence, despite some reservations (see J acoby et al., 1995), our results are best interpreted as evidence of the viability and promise of process dissociation in future research on diversity within spatial cognition. The significance of process dissociation for future research is further reinforced by our exploratory findings on correlations with self-reports. Some might argue that methods like process dissociation yield memory measures that are artificial and narrow, thereby failing to capture the holistic nature of spatial cognition. In contrast, weargue that such measures are necessary to understand the richness and complexity of spatial cognition. For both Experiments 2 and 3, estimates of conscious recollection from process dissociation were associated with participants self-reports of using configurations to aid their route learning. Our correlation findings also support the further argument that such measures are indeed reflective of the processes that are at work in coming to know and in remembering large-scale environments. Consistent with our conclusions, Schacter and Nadel (1991) reviewed evidence from neuropsychology and psychobiology that implicate diversity within spatial memory. In fact, they argue that research findings from cognitive psychology, neuropsychology, and psychobiology converge on a common theme-that several varieties or forms of spatial memory can be distinguished, and hence, spatial memory should not be viewed as a unitary or monolithic entity (p. 165). In neuropsychology, Schacter and Nadel reviewed evidence of selectivity in the effects of brain damage. For example, Landis, Cummings, Benson, and Palmer (1986) reported patients who had difficulty recognizing and finding their way through familiar environments but who were adept at using maps (e.g., locating the home on a map). In the context of our results, we suggest that examples like these reflect the dissociation between familiarity and conscious recollection. In concluding, we hope that experimental findings like ours will mark an important historical juncture in the evolving research tradition on spatial cognition. We see a research tradition that seemed to gain its initial momentum from recognition of the diverse things that people knew about environments and from recognition of the complexities of the methodological issues associated with measuring that knowledge (see Allen, 1985; Anooshian and Siegel, 1985; Newcombe, 1985; Siegel, 1981). Later, the focus seemed to shift from content to structure or organization (see Sherman and Lim, 1991), as exemplified by research 298 L. J. Anooshian and P. S. Seibert examining the hierarchical organization of spatial information (e.g., Allen, 198 1; Hirtle and J onides, 1985; McNamara, Hardy and Hirtle, 1989). Of course, such a focus on organization tends to be a focus on associations or integration rather than dissociations. Recent research findings have increasingly questioned the extent of hierarchical organization in spatial memory (e.g., Clayton and Chattin, 1989; Sherman and Lim, 1991). Similarly, it is likely that future researchers will increasingly recognize and explore the diversity of processes underlying spatial cognition (see also Schacter and Nadel, 1991). In appreciating environmental memory as a complex and multiply determined structure (Sherman and Lim, 1991, p. 292), future research efforts hopefully will reflect an appropriate balance between the search for integration and organization and the search for diversity and dissociations. REFERENCES Allen, G. L. (1981). A developmental perspective on the effects of subdividing macrospatial experience. Journal of Experimental Psychology: Human Learning and Memory, 7, 12&132. Allen, G. L. (1985). Strengthening weak links in the study of the development of spatial cognition. In R. Cohen (Ed.), The development of spatial cognition (pp. 301-321. Hillsdale, NJ : Erlbaum. Allen, G. L., Siegel, A. W. & Rosinski, R. R. (1978). The role of perceptual context in structuring spatial knowledge. Journal of Experimental Psychology: Human Learning and Memory, 4, 617630. Anooshian, L. J . (1988). Places versus procedures in spatial cognition: Alternative approaches to defining and remembering landmarks. British Journal of Developmental Psychology, 6, 38&39 I . Anooshian, L. J . & Kromer, M. K. (1986). Childrens spatial knowledge of their school campus. Developmental Psychology, 22, 854-860. Anooshian, L. J . & Seibert, P. S. (1996). Conscious and unconscious retrieval in picture recognition: A framework for exploring gender differences. Journal of Personality and Social Psychology: Personality Processes and Individual Differences. Anooshian, L. J . & Siegel, A. W. (1985). From cognitive to procedural mapping. In C. J . Brainerd and M. Pressley (Eds.), Basic processes in memory development: Progress in cognitive development research (pp. 47-101). New York: Springer-Verlag. Chown, E., Kaplan, S. & Kortenkamp, D. (1995). Prototypes, location, and associative networks (PLAN): Towards a unified theory of cognitive mapping. Cognitive Science, 19, 1-51. Clayton, K. & Chattin, D. (1989). Spatial and semantic priming effects in tests of spatial knowledge. Journal of Experimental Psychology: Learning, Memory, and Cognition, 15, Cornell, E. H., Heth, C. D. & Alberts, D. M. (1994). Place recognition and way finding by children and adults. Memory & Cognition, 22, 633-643. Evans, G. W. & Pezdek, K. (1980). Cognitive mapping: Knowledge of real-world distance and location information. Journal of Experimental Psychology: Human Learning and Memory, 6, Hirtle, S. C. & J onides, J . (1985). Evidence of hierarchies in cognitive maps. Memory & Hunt, M. E. (1984). Environmental learning without being there. Environment and Behavior, J acoby, L. L. (1991). A process dissociation framework: Separating automatic from intentional influences of memory. Journal of Memory and Cognition, 30, 5 13-541. J acoby, L. L. & Kelley, C. M. (1991). Unconscious influences of memory: Dissociations and automaticity. In D. Milner and M. Rugg (Eds), Neuropsychology of consciousness (pp. 201-233). San Diego, CA: Academic Press. 495-506. 13-24. Cognition, 13, 208-217. 16, 307-334. Diversity within Spatial Cognition 299 J acoby, L. L., Yonelinas, A. P. & J ennings, I. M. (1995). The relation between conscious and unconscious (automatic) influences: A declaration of independence. In J . Cohen and J . W. Schooler (Eds.), Scientific approaches to the question of consciousness. Hillsdale, NJ : Erlabum. Kirasic, K. C. & Mathes, E. A. (1990). Effects of different means for conveying environmental information on elderly adults spatial cognition and behavior. Environment and Behavior, 22, 591407. Kirasic, K. C., Siegel, A. W. & Allen, G. L. (1980). Developmental changes in recognition-in- context memory. Child Development, 51, 302-305. Kuipers, B. (1982). The map in the head metaphor. Environment and Behavior, 14,202-220. Landis, T., Cummings, J . L., Benson, F. & Palmer, P. (1986). Loss of topographic familiarity: An environmental agnosia. Archives of Neurology, 43, 132-136. LeCompte, D. C. (1995). Recollective experience in the revelation effect: Separating the contributions of recollection and familiarity. Memory & Cognition, 23, 324-334. McNamara, T. P., Hardy, J . K. & Hirtle, S. C. (1989). Subjective hierarchies in spatial memory. Journal of Experimental Psychology: Learning, Memory, and Cognition, 15, Mitchell, D. B. & Brown, A. S. (1988). Persistent repetition priming in picture naming and its dissociation from recognition memory. Journal of Experimental Psychology: Learning, Memory, and Cognition, 14, 213-222. Musen, G. & Treisman, A. (1990). Implicit and explicit memory for visual patterns. Journal of Experimental Psychology: Learning, Memory, and Cognition, 16, 127-137. Newcombe, N. (1985). Methods for the study of spatial cognition. In R. Cohen (Ed.), The development of spatial cognition (pp. 277-330). Hillsdale, NJ : Erlbaum. Presson, C. C. (1987a). The development of landmarks in spatial memory: The role of differential experience. Journal of Experimental Child Psychology, 44, 3 17-334. Presson, C. C. (1987b). The development of spatial cognition: Secondary uses of spatial information. In N. Eisenberg (Ed.), Contemporary topics in developmental psychology Presson, C. C., DeLange, N. S. & Hazelrigg, M. D. (1989). Orientation-specificity in spatial memory: What makes a path different from a map of the path? Journal of Experimental Psychology: Learning, Memory, and Cognition, 15, 887-897. Richardson-Klavehn, A. & Bjork, R. A. (1988). Measures of memory. Annual Review of Psychology, 39, 475-543. Schacter, D. L. (1987). Implicit memory: History and current status. Journal of Experimental Psychology: Learning, Memory, and Cognition, 13, 501-518. Schacter, D. L. & Nadel, L. (1991). Varieties of spatial memory: A problem for cognitive neuroscience. In R. G. Lister and H. J . Weingartner (Eds.), Perspectives on cognitive neuroscience (pp. 165185). New York: Oxford University Press. Seibert, P. S. & Anooshian, L. J . (1993). Indirect expression of preference in sketch maps of familiar environments. Environment and Behavior, 25, 607-624. Sherman, R. C. & Lim, K. M. (1991). Determinants of spatial priming in environmental memory. Memory & Cognition, 19, 283-292. Siegel, A. W. (1981). The externalization of cognitive maps by children and adults: In search of ways to ask better questions. In L. Liben, A. Patterson, and N. Newcombe (Eds.), Spatial representation and behavior across the life span (pp. 167-194). New York: Academic Press. Siegel, A. W. & White, S. H. (1975). The development of spatial representations of large-scale environments. In H. W. Reese (Ed.), Advances in child development and behavior (vol. 10, pp. 9-55). New York: Academic Press. Siegel, A. W., Kirasic, K. C. & Kail, R. V. (1978). Stalking the elusive cognitive map. In J . Altman and J. F. Wohlwill (Eds.), Human behavior and environment (vol. 3, pp. 223-258). New York: Plenum. Sloman, S. A., Hayman, C. A. G. , Ohta, N., Law, J . & Tulving, E. (1988). Forgetting in primed fragment completion. Journal of Experimental Psychology: Learning, Memory, and Cognition, 14, 223-239. Thorndyke, P. W. & Hayes-Roth, B. (1982). Differences in spatial knowledge acquired from maps and navigation. Cognitive Psychology, 14, 560-589. 21 1-227. (pp. 77-1 12).