I am interested in both perception and cognitive architecture. More specifically, I have been working in a number of areas related to conscious perception. My take on this issue is to identify areas in which the information which people process seems not to match the experiences people believe they have. For example, even though people can only see clearly in a small area in the center of their visual fields, they feel like they can see clearly over a much larger area. Similar examples appear to exist in backward masking (where it is possible to detect considerable processing of stimuli which observers claim not to have seen), attention (where people feel like they perceive a stable and comprehensive representation of the world, even though the type of information available changes with task and locus of attention) and even some linguistic effects (an example due to Andy Barss: Is “More people have been to Russia than I have,” a sentence? What does it mean?).
The goal is to understand why people’s beliefs about what they perceive or experience differs from what they can actually be shown to processes. It seems to me that such an understanding would have benefits of three kinds. First, it would have benefits for ergonomics and human factors. A fairly large percentage of accidents are caused in situations where people “looked for, but did not see” potential hazards. If we more accurately estimated how much we actually “saw” when we “looked” such accidents might be avoidable.
Second, understanding how people form beliefs about what they have seen is methodologically advantageous. This proved true in the case of cortical blindness where people were have been found capable of making judgments they do not believe they can make. My work suggests a similar phenomenon may be occurring with visual masking: people do better at making discriminations when forced to perform quickly. Thus slower responses may reflect non-perceptual effects such as memory, that also effect people’s beliefs about what they have seen.
Finally, understanding why people think they have particular conscious experiences may lead us to an understanding of what those conscious experiences really are. In fact it has been argued (Dennett, 1991) that the conscious experiences really are nothing beyond our beliefs that we have had them. Understanding the formation of these beliefs may give us insight into what has been one of the most contentious issues in both philosophy and psychology.
Visual Attention and Memory
Language and Cognition
"Masking" refers to a situation in which one stimulus (the mask), interferes with the perception of a second stimulus (the target). I am interested in such stimuli because subjects appear to process the stimuli despite denying that they have seen them. Is this a case of processing without conscious access? I have recently shown that for some such stimuli, the relevant variable appears to be the amount of time before the subject responds, rather than conscious access to the stimulus per se. The paradigm I used involves metacontrast masking, where a disk is masked by a subsequent ring which surrounds it. Observers were given the task of locating the masked disk. In those conditions where the masking was strongest, subjects were more accurate when they responded quickly than when they responded slowly. This fact suggests that perceptual information about the disk is initially available. The mask appears to act not by preventing the perception of the disk but by preventing it from being encoded into memory. If this result generalizes to related kinds of masking, it will fundamentally alter how we think about the phenomenon. What once invoked the complex issue of "unconscious perception" now looks more like a simple case of memory failure.
Related Publications and Presentations:
Lachter, J. 1996. The development of a percept. Invited presentation. University of Colorado, Boulder, Colorado, October 11, 1996.
Lachter, J. and F. Durgin. 1996. Metacontrast detection task shows anomalous speed-accuracy tradeoff. Annual meeting of the Psychonomic Society. Chicago, Illinois, October 31, 1996.(Abstract)
Lachter, J. and F. Durgin. 1997. Fleeting awareness in metacontrast. The first conference of the Association for the Scientific Study of Consciousness. Claremont, California, June 15 1997.(Abstract)
Lachter, J. and F. Durgin. 1997. Reduction in metacontrast with fast responding. Annual meeting of the Psychonomic Society. Philadelphia, Pennsylvania, November 20, 1997.(Abstract)
Lachter, J. and F. Durgin. In Press. Metacontrast masking functions: A question of speed? Journal of Experimental Psychology: Human Perception and Performance. (Abstract)
Lachter, J., F. Durgin and T. Washington. In Press. Disappearing percepts: Evidence for retention failure in metacontrast masking. Visual Cognition: Special Issue on Change Blindness.(Abstract)
If you look out your window you will probably see a complex scene of buildings, cars, pedestrians, trees and many other objects. How much processing does the visual system do on all of these objects? Or, as the question is put in the study of attention: Do we identify the unattended objects? There has been considerable debate on this issue. A number of studies have shown that subjects do identify stimuli which are irrelevant to their particular task. This fact is in accord with the intuition that we see much more than we can pay attention to. However, it seems to be in conflict with theoretical arguments that identifying many objects at once may be extremely difficult. In my thesis, I looked at two factors which might account for this apparent discrepancy. First, the factors which make the simultaneous processing of multiple stimuli difficult are compounded when the stimuli are in close spatial proximity. I found that subjects could process more stimuli in parallel when the stimuli were separated than when they overlapped, suggesting that the visual system may allow more parallel processing in situations were it is computationally tractable. Second, many of the experiments suggesting that subjects identify ignored stimuli present the stimuli in a way that might encourage sequential processing. I attempted to remove this confound. Under conditions which discourage sequential processing, I found evidence for processing of the ignored stimulus only when the subject was expecting a particular target. This suggests that one can search for a particular item in parallel, but can only identify one object at a time.
Related Publications and Presentations:
Lachter, J. 1994. Iconic memory for overlapping and non-overlapping stimuli. Converging Operations in the Study of Visual Selective Attention. University of Illinois, Urbana, Illinois, May 20, 1994.(Abstract)
Lachter, J. 1995. Processing of unattended objects in vision. TR 951 University of Rochester, Cognitive Science. (This is actually a reformated version of my disertation.) (Abstract)
Lachter, J. 1995. Processing of unattended stimuli: Detection not identification. Annual meeting of The Association for Research in Vision and Ophthalmology. Fort Lauderdale, Florida, May 15, 1995.
Lachter, J. Parallel processing of overlapping stimuli. Under revision.
When we look at a scene, we feel that we see the entire scene at once. We do not notice that our view is constantly changing despite the nearly constant motion of our eyes. This phenomenon has been explained a number ways. At one extreme, it has been argued that fixations are superimposed in a pictorial buffer. At the other extreme, it has been argued that we have no spatial memory aside from what can be encoded verbally. Together with Mary Hayhoe and Jerry Feldman, I examined this question using a paradigm in which observers were asked to classify angles as being acute or obtuse. In some conditions subjects could not see the points which defined the angle at the same time. Our findings suggest an intermediate position: people can remember the positions of a few points very accurately but fail to remember the positions of more than 4 or 5 points. Further, this seems to be independent of eye movements. Thus, it appears the spatial memory connecting our fixations is more precise than a verbal description, yet far less detailed than a pictorial buffer.
Related Publications and Presentations:
Lachter, J., M. Hayhoe and J. Feldman. 1989. Integrating form across saccades. Annual meeting of The Association for Research in Vision and Ophthalmology. Sarasota, Florida, May 2, 1989.
Lachter, J., M. Hayhoe and J. Feldman. 1991. Capacity limitations in the integration of information across saccades. Annual meeting of The Association for Research in Vision and Ophthalmology. Sarasota, Florida, May 1, 1991.
Hayhoe, M., J. Lachter and J. Feldman. 1991. Integration of form across saccadic eye movements. Perception 20:393-402.(Abstract)
Hayhoe, M., J. Lachter and P. Møller. 1992. Spatial memory and integration across saccadic eye movements. In K. Rayner (Ed.) Eye Movements and Visual Cognition: Scene Perception and Reading. New York: Springer.
Karn, K., J. Lachter, P. Møller and M. Hayhoe. 1994. Task dependent spatial memory across saccades. Behavioral and Brain Sciences 17:267-268.
Lachter, J. and M. Hayhoe. 1995. Capacity limitations in memory for visual locations. Perception 24:1427-1441.(Abstract)
Contextual cues are often important in identifying objects, especially if the object is viewed under unusual circumstances. However, little is known about how such cues are used. Similarly, in sentence comprehension the time course by which people develop a model of the information contained in a sentence is poorly understood. I am currently working with Merrill Garrett, Lise Menn, and Mary Peterson to determine if we can further our understanding of each of these issues by examining them together. We are studying the processing of sentences in which a picture has been substituted for one of the words (rebus sentences). Take the sentences "The fly was trapped under the glass," and "The fly was found in the glass." In the situation described in the first sentence, the glass is most likely upside down. Conversely, in the situation described in the second sentence the glass is most likely rightside up. Suppose a picture of a glass is substituted for the word glass in these sentences. Would an upside down glass be more easily identified in the first sentence than the second? If so, how much time must occur between the context and the picture-word before this effect can be detected? This work is still in the early pilot phase. However, we hope the answers to such questions will lead us to better understand how both sentence processing and vision interact with more general cognitive structures.
When I started graduate school my intention was to use connectionist techniques to study the interactions of attention, perception and memory. However, it became apparent that the simple feed-forward mechanisms that were the focus of most connectionist research at the time did not capture the interactions that were of interest to me. Further, many of the more interesting interactions that were displayed by these models were due to assumptions about the nature of the input that were unrealistic in the real world. Children must learn such things as where the boundaries between words fall and what sounds count as the same phoneme. By building such linguistic distinctions into models, connectionists are committing themselves to the highly nativist position, which they argue against. The paper I wrote paper with Tom Bever details this argument.
Related Publications:
Lachter, J. and T. G. Bever. 1988. The relation between linguistic structure and associative theories of language learning-A constructive critique of some connectionist learning models. Cognition 28:195-247 (Reprinted in S. Pinker and J. Mehler (Eds.) 1988. Connections and Symbols. Cambidge:MIT Press).(Abstract)
Lachter, J. 1994. Consciousness in natural language and motor learning. Behavioral and Brain Sciences 17:409-410.