Overview
Visual Orientation in Unfamiliar Gravito-Inertial Environments
Principal Investigator:
Charles M. Oman, Ph.D.
Organization:
Massachusetts Institute of Technology
Technical Summary
How do we know our location, orientation, and motion of our body with respect to the external environment? On Earth, gravity provides a convenient "down" cue. Large body rotations normally occur only in a horizontal plane. In space, the gravitational down cue is absent. When astronauts roll or pitch upside down, they must recognize where things are around them by a process of mental rotation which involves three dimensions, rather than just one. While working in unfamiliar situations they occasionally misinterpret visual cues and experience striking "visual reorientation illusions," in which the walls, ceiling, and floors of the spacecraft exchange subjective identities. VRIs cause disorientation, reaching errors, trigger attacks of space motion sickness, and potentially complicate emergency escape. MIR crewmembers report that 3-D relationships between modules - particularly those with different visual verticals - are difficult to visualize, and so navigating through the node that connects them is not instinctive. Crew members learn routes, but their apparent lack of survey knowledge is a concern should fire, power loss, or depressurization limit visibility. Anecdotally, experience in mockups, parabolic flight, neutral buoyancy and virtual reality (VR) simulators helps. Unfortunately, our understanding of how our sense of place and orientation is coded in three dimensions in the brain is incomplete.
The role of the NSBRI neurovestibular adaptation team is to do the critical experiments that provide a rationale and methodology for scientifically based countermeasures against inflight and postflight disorientation and motion sickness. The research spans Countermeasures Readiness levels one-six. Our specific project focuses on the role of visual cues in disorientation. Countermeasures under consideration or active development potentially generic and mission specific preflight visual orientation training using virtual reality and other simulation techniques. Also human factors standards for use of visual polarity and architectural symmetry cues, and the design of escape path signs, allocentric visual landmark systems, and you-are-here maps. Some of these are mature enough so that during the next year, we plan to begin working with the NASA JSC Countermeasures Evaluation and Validation Project to initiate non-advocate, evidence-based review and formal validation and implementation of some of our concepts. Meanwhile, a member of our research team (J. Richards) is spending three months at JSC this fall, working with Dr. Jon Clark, a flight surgeon who leads the JSC Neurological Function Integrated Project Team, to compile a more detailed and quantitative record concerning the actual operational incidence of visual orientation problems in the NASA-MIR and ISS programs.
The three major research themes (specific aims) of our project and the principal findings are:
1. Human visual orientation. (I. Howard, et al, York University). We have studied how visual cues determine spatial orientation and how ambiguous cues cause visual reorientation illusions. Using an eight foot "tumbling room" at York University, whose interior was furnished with tables, chairs, bookshelves, etc. we investigated the conditions in which the perception of self orientation with respect to the vertical is dominated by gravity, the visual frame of reference provided by the room's realistic interior, or by the principal axis of the subject's body. There is a natural tendency to perceive the feet as "down." It has long been known that moving visual scenes can produce compelling illusions of self motion (e.g. flight simulators and wide screen movie theaters), but it was not understood that motionless visual scenes could produce very large sensations of static tilt under certain circumstances. Howard and Hu (2000) showed that when gravitationally supine subjects viewed the interior of the furnished room that was similarly tilted 90 degrees with respect to gravity, so that it appeared upright with respect to their body, a majority of subjects reported they felt gravitationally upright. We call this a "Levitation Illusion." If subjects extended their limbs above their supine body, their limbs felt weightless. The strength of the illusion has been systematically studied in a large group of subjects with the room and the subject in all the different possible orientations, modulo 90 degrees. In certain other relative orientations, subjects experienced visual reorientation illusions - for example they perceived the floor of the room as a ceiling. We were surprised to see that susceptibility to the levitation illusion consistently increased with the age of the subject. Vestibular function is known to degrade with age and the association between the orientation of familiar visual objects and gravity (which we refer to as "visual polarity") is probably a learned phenomenon. In a related experiment, we also constructed a novel "mirror bed" device, which allowed us to quantify how the object property we refer to as "visual polarity" determines the strength of a VRI. A subject lying gravitationally supine in the bed views the laboratory through a mirror mounted at 45 degrees over his head. When strongly polarized objects are in view, the subject interprets the view as horizontal, and feels subjectively almost upright. The sensation of tilt is sufficiently compelling to produce visual-autonomic responses. When weakly polarized objects are seen, the subject feels nearly supine. Intermediate tilt perceptions can be created by manipulating the polarity (type and arrangement) of objects in the visual scene. Some objects (e.g. desks, chairs, saucers) have "intrinsic" polarity, because in daily life they are consistently seen in a specific orientation with respect to gravity. Other objects (e.g. blocks, pens, books) seem to have no intrinsic polarity until they are placed "on" another object. In this case, through their physical relationship the pair acquire what we refer to as "extrinsic" polarity. Understanding how the relative orientation of gravity, body axis and the visual scene interact is potentially very important for astronaut training, and also in entertainment and clinical applications. Strongly polarized objects and pictures may prove useful in reducing the incidence of disorienting VRIs in space station modules. Placing strongly polarized pictures in staircases might help some elderly people be less prone to falling.
2. Three dimensional spatial memory and learning. (C. Oman, et al, MIT and W. Shebilske, Wright State Univ.) On Earth, humans have a remarkable ability to keep track of their orientation and position relative to local landmarks. However, most large body rotations are made about the body axis which is aligned with the gravitational vertical. What are the limits of human ability to imagine, orient and navigate in a weightless environment, where one is free to turn completely upside down? MIR astronauts reported great difficulty visualizing the relative orientation of different modules on the station, and remaining oriented when traversing the central node module. Similar problems are anticipated on the ISS. Can spatial abilities in such 3-D environments be improved by preflight training? Most navigation and spatial memory research has addressed only the terrestrial situation. To find out, we have conducted a series of four experiments on human spatial memory. We designed a 3-D spatial task analogous to those confronting astronauts trying to learn the spatial relationships between the six entrance hatches in a space station node module of a space station. Subjects were placed in the center of a small six sided room. A picture of an easily recognized and remembered object was located on the center of each wall. Subjects had to memorize the relationship between the pictures, such that they could predict which picture would be where, even after the room was rotated about them into any one of 24 possible relative orientations. Subjects would be told their orientation relative to one of the walls, or shown the pictures on two of the walls, and then in darkness indicate the relative direction to a specific unseen picture. All six pictures would appear, and the subjects would have a few moments to study the relationships between the pictures before the next trial began. Some constraints were put on relative room orientation during the first two dozen trials to facilitate initial learning. However, by the time the subjects had completed sixty trials, most were able to reliably predict the direction of a specific target picture from any arbitrary orientation. The experiment was conducted using a head mounted display system to render a virtual room. However, it was repeated using a physical room, and very similar results were obtained. Tests also showed the gravitational orientation of the subject had little effect on the subject's ability to perform the task. However, the ability to do two and three dimensional mental figure rotations and recognize imbedded figures, as measured using conventional paper-and-pencil forms similar to those found on IQ tests, did consistently correlate with performance in each of our studies where we have tested it. Exit questionaires suggested our subjects chose to remember the relationships amongst the figures as they would appear with the room in a specific "baseline" orientation - often the orientation they encountered it at the beginning of the test. Many discovered they could memorize opposite pairs of objects, and learn the relationships between groups of three objects, from which the relative direction of all six could be inferred. Prior explanation and practice with the baseline orientation, pairs, and triads concepts tended to improve performance as compared to a control group, but we believe many of the control group subjects discovered these or similar techniques on their own. The best performers were the subjects with strong 2- and 3-D mental rotation scores. Many of this group said they were able to visualize the room interior, and mentally rotate themselves or the room, and make the correct judgement, only occasionally falling back on mnemonic rules like pairs and triads. To see whether subjects "learned how to learn" the task, in some of the experiments we trained subjects in two successive environments. As we hoped, they usually learned significantly faster in the second. In one experiment we brought subjects back in for retesting and found ability was retained one day, one week, and even one month after initial training. Another experiment showed that learning with randomly chosen rather than grouped (blocked) sets of room orientations enhanced ultimate performance. We are currently working on extending the paradigm to measure spatial memory across two previously learned modules, one of which is unseen. We want to know if coalignment of the baseline memorized module orientations is critical for performance. MIR crewmembers reported great difficulty visualizing the relative orientation of adjacent modules when the baseline orientations (established by equipment arrangements and visual verticals) were not coaligned. Our ultimate objective is to develop a methodology/pedagogy for generic and mission specific ISS preflight visual orientation training. Another application of this paradigm is in the design and evaluation of emergency escape route markings and systems of visual landmarks within modules that help crewmembers keep track of the principal axes of the ISS.
3. Neural coding of spatial orientation in an animal model. (J. Taube, et al, Dartmouth) Using a rat animal model, we have conducted experiments in our ground laboratories and in parabolic flight to better understand how our sense of place and direction is coded in three dimensions. In rats and primates, "head direction" cells have been found in the limbic system that appear to code head direction in a gravitational horizontal plane, independent of the animal's location, and roll or pitch of the head up to 90 degrees. The direction of maximum response ("preferred direction") lies in a fixed direction which varies from cell to cell. Under 1-G conditions, moving a prominent visual landmark around the animal results in a corresponding re-orientation of the preferred directions of all HD cells by the corresponding angle, a phenomenon that corresponds to the familiar human experience of emerging onto the street level from a subway, and reorienting your sense of direction based on viewing a familiar landmark. Until this project began, the response of HD cells had been studied only in a gravitationally horzontal plane. We have conducted a series of 1-G laboratory experiments in which rats are trained to crawl up a wall, across a ceiling (hanging upside down) and down the opposite wall, in an apparatus that allows us to verify the 3D response characteristics of HD cells, and infer whether the response sensitivity remains anchored by gravity to a horizontal plane, or whether the response coordinate frame of the cell re-orients to the plane on which the animal is locomoting. Results of these 1-G experiments indicate that cells in some animals continue to show robust direction specific firing in the same world-centered reference frame when the animal is walking upside down and response on the walls depends on the wall and whether the animal was going up or coming down, as expected. In other animals, the cells lose their direction specific firing on the ceiling. We suspect that some animals may be better at remaining oriented on the ceiling than others. In a separate series of experiments, we have also studied HD cell response chararacteristics in parabolic flight in a test chamber that was visually symmetrical in an up-down direction. All cells HD cells studied maintained their direction specific discharge when the animal was on the floor or the wall of the chamber. However, when placed on the ceiling of the chamber, HD cell directional specificity was frequently lost. However, in some cases, the preferred direction of HD cell response reversed across the visual axis of symmetry of the cage, as it would be expected to do if the cell's response coordinate frame reoriented to the ceiling. When humans roll inverted in parabolic flight and put their feet on the ceiling of the aircraft, they experience a visual reorientation illusion in which the ceiling seems like a "floor", and the left-right axis is reversed. We believe this is the first demonstration of the limbic correlate of a human 0-G spatial orientation illusion. Corroborating evidence has also recently come from space shuttle Neurolab experiments on limbic "place" cells which we believe are driven by HD cells (Knerim, et al 2000). Our results have suggested several important physiological questions, such as what mechanisms cause HD cells to lose their directional sensitivity? What kind of disoriented behaviors does it produce? Our experiments provide important insights on the role played by gravireceptors in stabilizing the human sense of place and direction not only in astronauts, but also in vestibular and Alzheimer's disease patients. Funding from NASA has allowed us to pursue the anchoring role of gravity as a major theme, and provided access to the unique facilities required for parabolic flight experiments. Involvement in the design of the experiments has helped all members of our project team form appreciate the close relationships between cognitive and cellular events.
It is never possible to predict the direction that an aggressive research program will take, but in retrospect we met or significantly surpassed our original goals in most areas, as defined in our 1997 proposal. Our discovery of 90 degree static 1-G visual reorientation illusions and their strong age dependency, the development of the mirror bed VRI technique, and its use as a research tool to quantify the strength of visual polarity cues and for vestibular autonomic research all were not anticipated. At the suggestion of advisory committees, our work on orientation and navigation has focussed on development and validation of our "virtual node" experimental paradigm, and comparison with a physical node. The physical node work was not originally proposed. However this focus has now led us to specific concepts for generic 3-D spatial memory training, and positioned us well for further research on 3-D navigation, and for evaluation of specific signs and landmarks. We have demonstrated a strong correlation between performance in our 3-D spatial memory task and measures of mental rotation ability, and not just field dependence, as we originally proposed. Our study of the effects of foot pressure cues is still in progress. Our development of an animal model for visual reorientation illusions in 1-G and 0-G has ultimately been successful, although it has proven more difficult than first anticipated to train the animals to crawl upside down across a gridded ceiling, so some of this work is still in progress. Another measure of our success is that there have been significant collaborations between investigators at the principal research sites and investigators on other teams. Dr. Oman participated in the design and performance of Dr. Taube's experiements; Dr. Shebilske designed the original "node" paradigm used by the MIT investigators. Dr. Howard and Dr. Oman have collaborated on the design of the VRI experiments, and Dr. Howard's mirror bed technique has been employed by Drs. Ramsdell and Wood of the cardiovascular team in their study of transient cardio-respiratory responses to visually induced tilt.