A Thesis Submitted In Partial Fulfillment of the Requirements For The Degree of Bachelor of Arts

   According to Franklin and Tversky's classic study (1990),  readers construct a mental model representing the spatial relations explicitly given in a text or narrative.  Their theory known as the spatial framework model states that the accessibility of objects depends on their position along three differentially accessible axes defined with respect to your body.  Therefore, readers will be fastest at making up/down decisions followed by right/left decisions, and lastly up/down decisions.  The purpose of this study was to replicate Franklin and Tverksy's classic study using a new narrative describing an office and determine whether their model would predict performance in a virtual environment.  It was hypothesized that the mental models derived from indirect sources such as text would be similar to those derived from direct visual experience using a virtual environment.  A total of 20 participants ranging in age from 18-22 were recruited from their Introductory Psychology classes to take part in this study.  Half of the participants were tested in the narrative condition and half of the participants were tested in the virtual environment condition.  Reaction times and error rates were recorded for all participantsÕ responses to factual statements regarding the location of objects within each condition (i.e., up/down; right/left; front/back).  The results showed that participants in the virtual reality condition were faster to respond to these statements compared to participants in the narrative condition.  Participants responded fasted to up/down statements, followed by front/back statements and finally right/left statements.  The results suggest that both indirect and direct experience produce similar mental models supporting the spatial frameworks theory originally proposed by Franklin and Tversky (1990).

   First, I would like to thank the Psychology Department as a whole for their understanding and support. I could not have accomplished this without you all, for everyone has assisted me in some way.  You have all realized that this has been the most difficult time that I have ever had to experience for various reasons, and your understanding has helped get me through it. Thank you Prof. Krachunas for your understanding and flexibility, and you, Prof. Flannery for your guidance, both with the study and with everything else. I would also like to thank you, Dad, for your unconditional love and support throughout my life, in good times and in bad. You have stood by me like no other, and I hope you are proud of this accomplishment. Finally, I would like to thank you Christalle, for being so brave through your ordeals. Your strength has given me the motivation to take action in my own life, and to do what needs to be done at any cost. Even in your darkest hour, you were selfless in the sense that you knew what I needed to do for my own future, and you stood patiently, waiting for the support that you so desperately needed. This is for you.

   The knowledge of space that an individual acquires throughout his/her lifetime is attained by both direct and indirect experience of the world. In everyday life, we hear or read about a particular event or situation through a number of sources. And from them, we form concepts of space from language. By hearing or reading language describing a place, object, or event, the recipient of the information can only indirectly experience it through anothers interpretation. Experience through language is an indirect source of spatial information, while visual, first hand experience is considered to be direct. Through direct experience, an individual constructs his/her own concepts and beliefs according to stimuli that is seen, rather than depicted. Because of this, memory of spatial information is believed to be encoded in memory through dissimilar cognitive processes, depending upon if the information was obtained through direct or indirect experience .
     To assist in memory acquisition of spatial information, individuals create cognitive maps, or mental models, used to code and simplify the way a spatial environment is arranged. When reading a description of a scene in which a person is surrounded by objects, people create spatial frameworks. Spatial frameworks are mental models that serve as a mental scaffolding (Bryant, 1997) on which information can be arranged and rearranged. The scaffolding is based on spatial concepts that define direction in terms of body axes and enduring physical and perceptual regularities to which the mind has adapted. The processes by which a spatial framework serves as a mental scaffold may not be well understood, but clearly they are determined by the nature of a real or virtual world. These frameworks are similarly used to remember observed physical scenes, either ones a person has actually been in or model scenes from the outside.
    Previous studies suggest that individuals may spontaneously use such mental models for verification and retrieval of spatial information at a later time.  Mental models may include information about spatial properties such as relative position and relative distance. According to Franklin and Tversky (1990), readers of written descriptions would retain more than a surface trace or a precise prepositional record of the text.  Furthermore, they propose that in order to keep track of the locations of objects in narratives, people form three-dimensional mental spatial frameworks based on their conceptions of space from interactions with the perceptual world. Researchers investigating memory for discourse have proposed that, in order to comprehend language derived from text, readers construct mental or situation models embodying spatial relations explicitly given in the text well as those inferable from the text. (Bransford, Barclay & Franks, 1972) Until 1992, few theories had been developed regarding spatial memory.
    In a study by Franklin and Tversky (1992) participants read narratives describing themselves in realistic three-dimensional environments in which they were periodically reoriented, and their response times for accessing information about objects in various directions with respect to themselves were measured.  Three classes of models making different predictions about behavior in this task were considered within the study to explain possible findings, however results only allowed for the spatial framework model.
    The Spatial Framework Model (Franklin & Tversky, 1990) reflects peoples conceptions of space based upon ones typical interactions with it.  They hypothesized that readers of their narratives would construct a three dimensional spatial framework, a mental model, or knowledge structure used to store, retrieve, and verify locations of objects relative to their own bodies. (Bryant & Tversky, 1992)  A spatial framework reflects the way people normally conceive of their perceptual world, based on their interactions with it.  This theory (Franklin & Tversky, 1990) states that the accessibility of objects depends on their position along three differentially accessible axes defined with respect to your body, including left-right, above-below, and front-back. Spatial frameworks predict accessibility of spatial relations from memory primarily on the basis of the relative asymmetry of the body axes, such that highly asymmetric axes lead to faster retrieval of information. (Bryant & Wright, 1995)  According to the spatial framework analysis, objects located on the vertical head/feet axis should be more accessible to the observer because they are both properties of the world and body. Also, this axis is physically asymmetric and normally correlated with gravity.  As observers navigate the world, vertical spatial relations among objects remain largely constant with respect to the observer whereas spatial relations in the horizontal plane change. (Bryant & Tversky, 1992) In other words, objects on the horizontal axis were not recalled as rapidly as those found on the vertical axis, because the left-right dimension is physically asymmetric and was not likely to have any gravitational significance to the observer.  Asymmetries due to gravity and constancy (Bryant & Tversky, 1992) under typical horizontal navigation render the vertical axis as dominant.  Finally, the front/back dimension is physically asymmetric, and the observer is perceptually and behaviorally oriented front-wards.  The objects that are located on this axis depend on the direction currently faced by the observer. According to the spatial framework model, the objects located on above/below axis would be the fastest to recall, because of its correlation with body asymmetry and gravity. Responses showed that the above/below axis allowed for the fastest recall, which suggests that subjects used a spatial framework in constructing mental representations to serve in memory.
    An important study by Bryant, Tversky and Franklin (1992) further investigated the spatial framework model. Similar to Franklin and Tverskys pioneering study, subjects read narratives and were then probed for the locations of objects.  Each narrative was described from one of two perspectives, an internal perspective of an observer within the scene, surrounded by objects, or an external perspective of an observer outside the scene, with objects in front.  Results showed that for the internal spatial framework, readers were faster to questions of front or back, reflecting the perceptual and biological asymmetries that favor an observers front. (Bryant & Tversky, 1992) For the external spatial framework, all objects were in front of the observer and readers were equally fast to questions front and behind. They concluded that the difference between internal and external spatial framework reflects the different perceptual experience of observers in the two perspectives.
     More recent studies have continued to test various hypotheses regarding spatial memory within dynamic scenes or environments, rather than static configurations.  A possible reason for this is due to the idea that three- dimensional environments simulate experience and encourage participants to use mental models. technological advances have allowed researchers to conduct more complex studies, which require a more life like representation of a particular environment. The most highly used advancement in this area of experimentation is virtual reality. Virtual reality is a computer simulated three-dimensional environment that people can interact with and explore in real time. It has been used for various studies within many disciplines, and it has been found to be especially useful in experimental psychology.  Recently, there has been a growth of interest in VEs as a tool for investigating spatial learning. This study calls for the creation of an environment, which contains various objects in an array. Rather than isolating a particular setting, and keeping the locations of the objects constant, virtual reality was a welcomed experimental device, because it permits the creation of environments of varying complexity and allows interactive navigation. (Belinguard & Peruch, 2000)  Virtual reality is beneficial because manipulating the structure of a real environment is difficult or even impossible. The potential benefits of VEs as training media for optimizing environment behavior interactions have been accepted for many years; for example, in flight simulation and battlefield training (Wilson, 1999). The two most popular ways to present a virtual environment are via a head-mounted display, which updates the image in synchrony with movements of the users head, and via the conventional computer monitor or large screen projection system.
    There is evidence of substantial similarities in the spatial knowledge that is acquired in real and virtual environments. Wilson (1999) claims that spatial information is learned using the same cognitive processes in both real and virtual environments. He had concluded that one factor that may promote learning in both conditions is the observers active involvement in the exploration. Certain factors were found to influence the active-passive dichotomy, such as attention. A lack of attention while learning can effect ones ability to recall information. The environments that are created for studies of this nature are usually simulations of actual settings in which the observer is most likely accustomed to, except what is actually being seen are computer graphics, which seem to be lifelike, due to their three-dimensional appearance. However, because the possible confound of attention, we have used a rather interesting environment, in hopes of keeping the interest of the participant.
    In the present study, half of the subjects experienced the environment according to how it was described in a narrative.  The findings of the previously mentioned study suggest that an active participant is more likely to have developed a more detailed mental representation of the environment.  With this idea, it could be hypothesized that those subjects within the present study who read the narratives would have more rapid, accurate responses than those who passively experienced the same setting at the control of the experimenter.
    A goal within the present study is to compare spatial memory among subjects whom have read a description of an environment with subjects whom have experienced that very same environment.  Prior evidence suggests that spatial mental models derived from text are similar to those derived from direct experience, however we hypothesize that this claim is consistent within a virtual environment.  Similar to a study done by Franklin & Tversky, this task was designed to capitalize on peoples life long experience of encountering objects while navigating in complex environments. (Franklin & Tversky, 1990).  Results may support the idea that the spatial framework model also applies how individuals encode the locations of objects while navigating through an environment, rather than reading a description of the very same setting. The research reported here investigates whether the theories proposed by previous studies will be consistent in comparing the recall of objects within a virtual environment versus reading a narrative of the same setting. Presenting a three-dimensional environment with virtual reality is used here to retest if the spatial framework theory is consistent within that condition.
    The question of interest is if the responses of an observer of the virtual environment will form create mental models according to the spatial framework model, or any other theories that were proposed within the investigations of Tversky, Franklin, and Bryant. In comparing recall in both conditions, results may show that different concepts in mental imagery may govern memory and perception of space. This study attempts to demonstrate that the mental spatial framework model is employed within visuo-spatial memory, as it has been found as the explanation in how individuals encode spatial information from language/text.

Participants

A total of 20 (10 male, 10 female) participants from a small, liberal arts college in the northeast were recruited for this study. Each participant received credit for their introductory psychology course upon completing this study. Participants ranged in age from 18 to 20, with the average age = 18.0 years. All participants provided full consent prior to testing.

Materials

Narrative: A printed narrative was created that described an environment containing various objects. The story was written in the second person from an internal viewpoint, meaning the observer was immersed within the setting, rather than viewing it from a distance. The objects were common to the environment being described. The narrative was comprised of descriptive sentences, which oriented the reader toward a particular object. Also, a number of filler sentences were placed within the text. The narrative totaled 358 words, which was printed in courier new, bold lettering, then laminated. (See Appendix A.)

Virtual Environment: A personal computer was used for the virtual reality portion of the study. Superscape, a virtual world development program, was used to create the environment. The depicted environment was a rectangular shaped room, with the walls to the left and right being longer than the front and back wall. The room contains familiar objects in various locations on one of two levels. (See Appendix B.) for views 1-4)

Mental Model Task: The mental model task was created using Superlab LT. Using this program, a total of nine statements were presented upon a computer screen in random succession, and all represented a particular object that was described in a given location within the environment. All statements described the possible location of an object in relation to the participant. All three axis, left-right, above-below, and front-back were equally represented. There was intentionally no practice trial, which will be explained in greater detail within the discussion. This program was used to record and measure response rate and accuracy for all 9 statements. Responses were measured in milliseconds and accuracy was recorded as correct (C) or error (E). (See Appendix C for probe statements)

Procedure:  In the first experiment, half of the participants were verbally instructed to read the narrative, and then were informed they would be answering a series of questions regarding the story. They were told to read the story only once, at whatever pace they were most comfortable. When the participant had completed reading the story he/she was instructed that they would not be able to refer back to the story for the next phase of testing. Upon completion, the participant was presented with a number of statements regarding the narrative on a computer screen. They were requested to evaluate whether the statement was true or false, according to how an object was presented within the narrative. Furthermore, the participant was to press the 1 key if the statement was true, and the 2 key if the statement was false. Subjects were told to respond to each statement at their own pace, without the opportunity to refer to previous statements. Subjects were unaware that response times and accuracy were being recorded and measured.
    The computer presented one sentence at a time, and the participant advanced to the next statement by pressing either response key, regardless of accuracy.  During each trial, no feedback from the experimenter was given.
In the second experiment, half of the participants  were tested using a virtual environment. In our testing protocol, the experimenter moved the participant along the same pathway as described in the narrative. The virtual room was identical to the setting illustrated within the narrative. The experimenter led the subject slowly around the room once. Exploration time (approximately two minutes) was intended to be similar to that of the duration of time in which it would take an average skilled reader to complete the narrative. This was measured by timing a group of 8 random participants, who did not take part in the core study.  Upon completion of the exploration, the subject was presented with the same statements as those who had read the narrative. Identical verbal instructions from the narrative portion of the experiment were given to those within this phase of the study.
All participants were fully debriefed regarding the specific goals set forth for this study.

    A 2 Condition (Virtual Reality; Narrative) by 3 Question Type (Up/Down; Right/Left; Front/Back) Analysis of Variance (ANOVA) was computed for reaction time scores.  A significant main effect for condition was observed, F(1, 18)=4.84, p<.05. Reaction times were faster for participants in the virtual reality group (M=5852.67, SE=703.63 compared to the control group (M=3661.79, SE=703.63), irrespective of question type.  There was also a significant main effect for question type, F(2, 36)=9.05, p<.001.  Planned comparisons (dependent t tests) showed that reaction times were significantly faster (p<.05) for up/down (M=3449.00, SE=489.71) versus right/left (M=6240.52, SE=809.62).  In addition, reaction times were significantly faster (p<.05) for front/back (M=4582.17, SE=655.03) versus right/left (M=6240.52, SE=809.62).  There was no significant difference between up/down and front/back reaction times.  Finally, there was not a significant interaction between condition and question type for reaction times.  Note, the same pattern of results for reaction time was observed when restricting the data analysis to correct responses.

    A 2 Condition (Virtual Reality; Narrative) by 3 Question Type (Up/Down; Right/Left; Front/Back) Analysis of Variance (ANOVA) was also computed for error scores.  There were no significant effects to report for this analysis.  This may be due to the fact that error rates were relatively low for each question (e.g., up/down 1=25%, up/down 2=10%, up/down 3=10%; right/left 1=25%, right/left 2=25%, right/left 3=30%; front/back 1=10%, front/back 2=25%, front/back 3=25%).

    The results of this experiment suggest that both indirect and direct experience produce similar mental models, supporting the spatial framework theory originally proposed by Franklin and Tversky (1990). However, the absence of correlation between response times in both conditions does not allow us to draw any conclusions about the effect of language versus visuo-spatial learning in spatial memory.
    Overall, the predicted pattern of response times, based on asymmetries of the human body axes and the correlation of the above/below axis with gravity, were similar within both experimental conditions. As in Franklin and Tversky's study, the above/below dimension was the fastest in object location recall, thus supporting the notion that the subjects encoded spatial information according to the spatial-framework model. The data also disproved the idea that subjects would have more rapid response rates in the virtual reality phase of testing, because they experienced the environment directly. It was found that active, or direct experience through exploration of a virtual environment does not promote memory for information regarding objects and their location.
    A possible reason for lack significant findings could be that participants who had experienced the virtual reality were led around the environment just as described within the narrative. We believe that response times and accuracy rates would have differed significantly if the participants were instructed to actively explore the room with no guidance from the experimenter. Most likely, they would have taken a different route through the environment, and therefore objects would be located differently in relation to themselves as described in the text. For example, if the participant (non-guided exploration) had began the navigation by walking to the right of the room, up the stairs, and to the left, the grandfather clock would be to the persons right, instead of the left as described in the narrative. If in future research, a subject is instructed to actively explore an environment, then the narrative must be modified by describing objects in relation to other objects, rather than to the observer. The reason for this is simple; the locations of objects in relation to the observer change as he/she moves within the environment. By describing the objects according to their relationship (position/distance), their locations will be consistent within both conditions. A problem with this alteration is that it will not test the spatial framework model, because objects are not described relative to a body axis, so therefore if the testing conditions were modified, so would the hypothesis. The spatial framework model is a mental model, or knowledge structure used to store, retrieve, and verify locations of objects relative to their own bodies, not other objects. A new study could involve a narrative that describes the location of an object(s) according to its position in relation to another object, as well as participants freely exploring a virtual environment. The testing procedure would be similar to the present study, however, the statements would illustrate the location of an object not in relation to the observers position.
    In our research, we began with the premise that participants will employ the spatial framework theory while learning the locations of objects while actively and directly exploring a virtual environment. It was surprising that participants on the average had similar response times within both conditions. Our assumption was that reaction times would be faster for those within the virtual environment condition, because of direct experience of the same environment as depicted within the narrative.
    Data did suggest that participants did employ the spatial framework model in both conditions, because as hypothesized, reaction times would be fastest for the above/below axes as previously demonstrated in studies by Franklin and Tversky (1990,1992). Subjects were able to form an understanding of physical properties such as asymmetry and gravity by reading a description and through active exploration through imagery, hence the basis of the mental spatial framework theory. All other theories were considered in analyzing data, however our findings support the spatial framework theory in both conditions within the experiment.
    Because there was no significant differences between reaction times for both the narrative and the virtual reality conditions, this leaves very little to be said.  If there was indeed a significant correlation, some possible issues to be raised would include how the virtual environment contained more stimuli that was not described in the narrative, which would flood observers with unnecessary information. Many possibilities exist, however, since there was no correlation, these possibilities are irrelevant to our findings.

    Overall, our results were similar to the classic study by Franklin and Tversky. We are able to further conclude from this experiment that participants employed a spatial framework while encoding spatial information in reading a description and through actual exploration of a virtual environment. More research is needed to further exemplify how mental spatial frameworks are used in mental imagery for virtual environments versus narration/language. More specifically, it is important that we determine whether the acquisition and recall of spatial information is faster through direct experience. Once this has been determined, we may be able to enhance the way in which we learn and store spatial information. Enhancing spatial ability would greatly benefit our performance in everyday tasks, at work, at home, on the field, or in the classroom.

THANK YOU  FOR VISITING MY WEB PAGE.