Lecture Notes for March 2-30th

We began by looking at the Eriksson article. She reviews the history of tactile pictures and begins with a discussion of the meaning of terms such as "picture" and "image".

Eriksson notes that there is a history of tactile pictures being accompanied by text, i.e., there seems to have been an assumption that tactile pictures require a linguist context. After her discussion of the history Eriksson moves on to a discussion of the perception and understanding of pictures. She quickly reveals a Helmholtzian perspective and assumptions about the relations between tactile pictures and mental imagery. We discussed this as a class.

Eriksson then makes the claim that the elements of a picture must occupy the same positions in the depicted object as in the real one. We discussed this problematic claim and realized that it is inadequate because it is not always true. We noted the "same position" cannot mean exactly the same nor can it mean the same set of relations between elements of the pictures and features of the depicted object. So, the claim degenerates into the claim that there must be some similarity in the sets of relations. We then discussed similarity and how this was no longer a helpful explanation, if an explanation at all.

Eriksson then moves on to discuss the process of translating visual pictures into tactile pictures. She discusses a process of interpretation, but we noted that this is ambiguous because it is not clear what is the result of convention and what is the result of necessity. She notes that certain kinds of pictures have to undergo significant even radical transformations while others require little or no such transformations.

Interestingly the pictures that require the least amount of transformation are spatial pictures. So, it seems that spatial information can be transferred without difficulty. It is not clear what the factor is that makes other pictures more difficult to transfer. For Eriksson it seems to be how complicated the picture is. But complexity and simplicity are notoriously difficult to define. There may be a relevant point here but it is difficult to get clear about it.

When Eriksson tries to clear this about by comparing the efficiency of touch and sight she seems to confuse perceiving patterns and sensing sensations, i.e., her Helmholtzian framework may be the source of confusion at this point in her discussion. We discussed this in connections with quotes from Eriksson. We noted how important the application of critical reflection and theoretical analysis is to the reading of such articles.

We then moved on to discuss the Pring and Eardley article. The article is a summary of the research concerned with the mental representations that individuals with/without visual impairment bring to the experience of art.

They reviewed the literature concerning language development and we noted the research that shows that blind children can use visual words correctly. Their ability to identify emotions, the mental states of others, and social understanding is delayed, but they do catch up.

Pring and Eardley then review the literature on object recognition. The research indicated that those with and without sight develop "broadly similar structural descriptions of objects." We noted the difficulty of using similarity, but then we moved on to discuss structural descriptions in which the relations between parts and wholes is represented. We noted the relevance of this since structure often predicts function.

Pring and Eardley then discuss the need for predictive touch which we discussed as a valuable application of the Helmholtzian tradition. In connection with this Pring and Eardley discuss a Pring and Walker 1993 study that showed that if tactile pictures were accompanied by a verbal label then blindfolded children and children with visual impairment have "little problem recognizing the structural components of a picture."

We then discussed how top-down processing interacts with bottom-up processing in all of perception. And this may explain the value of accompanying text that Eriksson noted in her historical review. We noted that this top-down information does not have to be verbal in nature. McGinnis and Pring (1998) study indicated that blind subjects could pick up stylistic variations and use this to successfully categorize pictures.

Pring and Eardley then move on to discuss the role of imagery and imagination. They review the work of Kosslyn and others that showed that the primary visual cortex is activated during imagery tasks. Kosslyn took this as definitive proof for the internal picture theory of mental imagery. However, more recent indicates that in people who have blind since birth parts of the PVC have been hijacked for tactile tasks including spatial imagery tasks. Recent work has also shown that sighted people also use the PVC in tactile tasks. So the conclusion that imagery is internal vision is actually not supported by more recent work. Imagery seems to be much more spatial than visual.

Research in the 1980s indicated that the congenitally blind can use imagery effectively to preserve metric spatial information. More recent research indicated that as the memory load is increased their performance degrades. However, this deficit can be significantly reduced through practice which suggests that the deficit is an encoding issue rather than an inherent defect of spatial information acquired through tactile/haptic means.

We then moved on to discuss the Heller article. Heller begins by making an important argument. He notes that blind people sometimes have difficulty identifying pictures of unfamiliar objects. However, the same is true of sighted people looking at pictures of objects from an unfamiliar point of view. We do not conclude from the failure of the sighted that vision is not suited to perceiving pictures. So by parity of argument we should not conclude that the failure of the blind indicates that touch is inadequate for perceiving pictures.

Heller has found the supplying people with a category helps them identify tactile pictures. Similarly performance is excellent if individuals are told to the picture of some object among a set of pictures. This again demonstrates the important role of top-down processing and such the relevance of the Helmholtzian (perhaps Kantian) framework.

Heller then performed an experiment comparing the blind and blindfolded sighted people with raised line drawing of the water level problem. The congenitally blind performed as well as the blindfolded sighted subjects.

Heller also found a group of low vision subjects using only touch outperformed college students using sight. So, the blind are capable of very sophisticated perceptual problem solving using tactile pictures.

We then discussed Heller’s adjustable board experiment. The central problem was whether the blind use foreshortening to represent changes in depth and slant. Heller found that the did not spontaneously do so.

However, Heller makes use of his previous argument and notes that many sighted people have to learn to draw in perspective and make use of foreshortening. They often find this very difficult. In short, a performance error such as drawing is not conclusive evidence of a competence defect.

We then discussed the competence/performance distinction. Heller wonder if the blind have the representational competence even thought they demonstrated drawing (performance) failures. He explained how the tactile pictures showed how the board appears to the sighted as it is moved through various positions. This is all that he explained. Nor did he pair the pictures with the board positions. He then had subjects feel the board and find the matching picture. The congenitally blind performed very well on this task with some of the subjects making all the correct assignments. So, it looks like the congenitally blind do have the relevant representational competence.

We then discussed the Jannson and Holmes article. They begin by noting that a viewer can perceive a picture as two things at the same time, a 2D object and a 3D scene. They claim that haptics does not have the ability to perceive 3D aspects of pictures. In fact, it is widespread opinion that 3D perception in pictures with haptics is almost impossible.

They note that Kennedy (1993) obtained evidence that the blind appreciate basic principles of perspective." (See also the more recent article by Kennedy on Tracy.) They also note that Heller et. al. (1996) found evidence that the congenitally blind were able to discover the meaning of perspective in tactile pictures without feedback information.

So, Jansson and Holmes discuss a point derived from Gibson, viz., the use of texture gradients to specify depth. Texture gradients follow very stable rules for depth, e.g. the farther the units are away the smaller they are and the close they are together. This is a very powerful depth indicator.

So, they ran a series of experiments to see if texture gradients on their own could provide depth information to the blind. They also wanted to see what happens as more information about tactile texture gradients is provided to subjects. The results indicate that the blind cannot do this spontaneously and that they need significant information in order to make use of texture gradients. This seems to be significantly different from visual perception. However, we did note that the experiment was performed on blindfolded sighted subjects. And this may be a confound. A similar experiment was performed on the early blind and the results appear to be similar.

We then discussed the Richardson article. Richardson addressed the question as to how tactile pictures are best explored. Richardson points out that the evidence on which form of touch is best is very mixed. Richardson then points out a possible confound, viz., many studies involve the subject only using one finger. How ecologically valid is this?

Lederman and Klatz (93) obtained evidence that people chose to use one finger to examine 3D objects under ecologically valid conditions. Symmons and Richardson (2000) wondered if the same was the case for 2D tactile pictures. Once again blindfolded sighted subjects were used so we must be cautious. Nevertheless, their study indicates that one fingered exploration is very ecologically valid. They conclude that they have addressed this confound issue and that their results support studies that restrict exploration to one finger. They do caution that their study needs to be replicated with blind subjects. This is an example of very careful and very good science.

Richardson then notes that Gibson (1962) argued that active touch is superior to passive touch. Yet the empirical research on this topic has been mixed. Richardson then argues that this is due to the presence of confounding variables due to the diverse testing conditions without standardized experimental design. Richardson correctly argues that a significant difficulty is that there is no agreement on what is meant by "active" and "passive." Gibson’s touching versus being touched is very inadequate. So Richardson makes use of more careful conceptual analysis that clarified this issue by taking into account factors such as relative contribution of kinesthetic and cutaneous inputs, methods of gaining information, and the presence of absence of volitional control. Symmons and Richardson developed the Tactile Display system to control for these variables. Using it they obtained evidence for the following conclusions.

Then subjects had to identify simple continuous outline drawings with little or no internal detail then passive touch is better. When the task was to distinguish among shapes then there was no difference between active and passive touch. When the stimuli we not continuous but contained junctions or internal detail then active touch is superior. Richardson speculates that passive touch’s superiority may be due to the fact that the memory load is lower. This is an interesting proposal but it does not explain why the two forms of touch are equal in the discrimination task, nor why active touch is superior under the more difficult condition of non-continuous pictures with significant internal detail.

Vervaeke suggested that we take a cue from the problem solving literature. Think of the tactile tasks as problems to be solved. Often a problem requires restructuring in order to solve it. Often breaking out of an inappropriate formulation of the problem and have some insight into a superior formulation is key to solving a problem. Such restructuring requires the redirection of attention and the creative reframing of information. This would become more relevant as the problem becomes more complex. So, active touch is superior in these kinds of conditions in which effective formulation and restructuring are key precisely because it redirects attention and leads to a more creative involvement with the information.

For the final class we looked at two philosophers on the issue of pictures and the blind. Lopes argues that work with pictures and the blind may significantly alter our conception of the hierarchy of the senses and what a pictures is and what its significant aesthetic properties are. This is in contrast to Hopkins who argues towards the opposite conclusion. Hopkins argument is sophisticated and we need to consider it carefully. Hopkin’s argument centers on the claim that the aesthetic properties of a picture, insofar as it is a picture, are dependent on the representational properties of a picture. He then argues that tactile pictures lack significant representational properties found in visual pictures, and therefore must have different aesthetic properties.

He notes that there are two prima facie responses to the issue of tactile pictures. The first is that there are no significant representational properties between tactile and visual pictures and therefore no significant aesthetic differences. The second is to claim that there are significant representational differences and therefore the aesthetics would be different. These two claims conflict. One can keep the first claim and abandon the second. In contrast, one can keep the second and give up the first claim which is what Hopkins does. In order to do this one needs to find a relevant and important representational difference between tactile and visual pictures. AND one needs to give an alternative explanation of what is going on in tactile pictures that preserves the uniqueness of visual pictures. Hopkins attempts both of these tasks.

In order to tackle the first task Hopkins makes use of Reid’s concept of an object’s visible figure. According to Hopkins "visible figure is a matter of the directions, from some point, in which parts of a thing lie" Reid argued that a picture of an object reproduces the set of directions of an objects parts from a particular point of view. This is the way in which a picture resembles, and therefore represents, its depicted object.

We’ve seen something like this before with Eriksson and it appears that the same objection applies here as before. A picture and its objects need not have the same visible figure. For example, a picture lying in the horizontal plane cannot have the same visible figure as the vertical real world depicted object. The set of directions would just be different. Yet such pictures are perfectly comprehensible. So the claim about the nature of depiction is questionable. This is important because Hopkin’s argument is that visible figure is only visible and therefore cannot be found in tactile pictures. So, Hopkins argues that pictures work in terms of visible figure and visible figure is only found in visible pictures, and therefore pictures cannot be tactile. We’ve seen that the first claim about how pictures work is questionable, but what is the argument for the second claim? He argues that in vision we have the convergence of direction for parts of an object the more distant those parts are. However, Hopkins argues, this is not the case when we touch the edges of an object. Even as our hands move into the distance they remain at the same distance from each other.

He also argues that visible figure is relative to a point of origin from which we are considering an object, and no such point exists for touch.

The first argument is flawed because it seems to be restricting touch to a single hand touching a single point. However, if I use two hand to touch the same point and then move that point into the distance then one notices that my arms go through significant convergence. So if I’m focusing my hands and the same points I definitely get convergence effects for distance.

What about the claim that there is no point of origin for touch? Hopkins argues that no one point on the hand has priority when we are touching an object. So there is no point of origin. What about the point of origin being the location from which someone is touching? Hopkins argues that the point of origin for vision is quite precise and therefore can produce good pictures while spatial location is a very imprecise point of origin and therefore will produce very poor pictures.

His point seems to be that the point of origin for touch would be the chest or torso (p. 195), and this is too large for the high definition necessary to match visible pictures. And there does not seem to be any privileged point on the chest. However, we discussed the possibility that touch could uses the intersection point between the center and the two lines of the two arms. The center axis is very important to movement and to deciding direction even direction on the chest, e.g. left and right sides of the chest.

So there is good reason to believe that active touch could have a fairly precise point of origin. Hopkins seems to be biased towards a very static conception of touch.

Note his response to Lopes point about a blind person in Paris must have a cognitive map with various points of origin and figures of the city. Hopkins’ reply is that the person is making use of information other than touch. He mentions acoustic information, movement information, and kinesthetic information. The second and third a factors of active touch! Hopkins has also admitted that visible (spatial?) figure can be formed without vision, and therefore vision is not a necessary condition.

What about tactile pictures? Hopkins replies that we need to make a distinction between experience and belief so while a picture can cause to experience shape it can only cause us to beliefs about heat. In the same way, tactile pictures create beliefs about visible figure but they cannot produce the experience. We’ve seen that there is good reason to doubt these claims about visible figure. If visible figure is actual spatial in nature then there is no reason that tactile picture could not create the experience of spatial figure.

NOTES FOR THE KENNEDY PAPERS
Make sure you have a basic grasp of what happened in the studies. Make sure you can point out how it is relevant to issues we’ve discussed. For example, Tracy’s work on perspective seriously challenges Hopkins’ point about visible figure. Do not work about details of experimental design. Understand the main arguments and points and how they reflect upon issues discussed in the course.