With this essay I want to rejoin a discussion in which I first participated ten years ago, on the question of non-human — particularly machine — agency. My renewed interest in questions of agency is inspired by developments over the last decade both in the area of interactive computing and in the debate over agency within social studies of science and technology. What I propose to do is another attempt at working these two projects together, in what I hope will be a new and useful way. The newness is not so much a radical shift in how we construct human-machine boundaries, as a reconsideration of those boundaries based in critical reflection on my own previous position, in light of what has happened since. What follows is the beginnings of the argument.
A series of experiments was conducted to explore sources of difficulties in manipulating daily electronic appliances. In realizing users' goals, these appliances require users to decompose their goals to a set of subtasks in a specific way. We hypothesize that users who have extreme difficulties do not decompose the task or that their decompositions are different from the one that designers assume. To test this hypothesis, we compared the performance of those who had been taught the general idea of task decompsition with that of the control group, using a copy machine. The results of the two experiments showed that the trained group achieved the tasks faster with fewer errors than the control group. The experiment 3 examined this task decomposition hypothesis, using a refined copier that visualized the idea of task decompostion. The results showed that the refined copier reduced errors at the early stages of learning. These results strongly support the idea that users' difficulties lie mainly in the task decomposition.
Human-Computer interaction may be correlated with social interaction and cultural norms. This study explores the validity of the hypothesis that Human-Computer interaction displays the same dynamics as Human-Human interaction. We examined how Japanese and American people respond to a computer as social entity and how cultural differences between Japanese and American behavioral norms in reciprocal social interaction play a role in this response. We first conducted an experiment to test the hypothesis that people interact with a computer as they would with another human. Two groups of subjects experienced positive/negative interaction with a computer which gave them Helpful/Unhelpful tips in completing a Desert Survival task. Subjects were then requested to help either the same computer or a different computer by supplying it with data in a Human Color Perception Task. The results showed that (1) both Japanese and American subjects exhibit behaviors in accordance with social norms of reciprocity, but (2) there is considerable difference between Japanese and American subjects in some conditions, which suggests cultural dependence in people's reaction to computers. In the second experiment, Japanese subjects were tested, after a positive interaction with a computer, in the following three conditions: subjects were requested to provide information to (1) the same benefactor computer, (2) a new computer that belongs to the same group as the first benefactor computer, or (3) a new computer which belongs to a different group from the benefactor computer. The results showed that reciprocal behavior in Japanese subjects is group-oriented, which presumably reflects the cultural characteristics of the Japanese in Human-Human interaction.
Sixty pairs of University students participated in referential communication tasks with Tangram figures. Using referential communication paradigm, the present study aimed at investigating the effects of cognitive artifacts employed in the conversation. The members of the pairs, who were familiar with each other, were randomly assigned to either a director or a matcher. The task for the director was to give a information to her/his matcher so that she/he could identify 8 Tangram figures in order. The six trials were given in different order to the pairs. The pairs were randomly assigned to the one of 4 communication channel conditions; same room separated by screen, or different rooms using a telephone with 3 levels of interference noise (none, low and high). Even though task performance was not affected by communication channel conditions, there were substantial differences in the variety of communicative tools employed in the dialogues, especially in the first trials. Qualitative and quantitative analyses of the conversations showed; (a) in the screen and no interference telephone conditions, participants frequently described the figures in detail and with reference to geometrical forms, but not for the telephone interference conditions, (b) participants in the low interference telephone condition employed more metaphoric expressions for small parts of the figures, whereas the participants in the high interference condition tended to use metaphors for the whole, a strategy which was effective when the pairs shared common ground, but less so when this common ground was missing, (c) in the high interference condition, the number of reference units was small in the first utterances, but increased in response to the matcher's feedback. The three kinds of feedback from the matcher (requests for repetition, clarifications, and questions) illustrated the cooperative nature of constructing common ground in conversation. The results show that changes in conversation due to interference can be explained in terms of reduced attention and/or changes in strategies used in making conversation.
It is well known that the arrangement of humans and objects plays an important role in everyday human communication. Thus it is conceivable that in video-mediated communication such importance will be preserved: especially when it comes to the spatial arrangement of body parts. We used this observation as a hypothesis to implement a remote collaboration system. For this paper, the optimal arrangement of communication resources for an education system to support collaborative learning, including hand-manipulation of physical objects was empirically investigated. The authors propose a ‘body metaphor’ concept as an initial design guideline. It follows the ordinary arrangement of human bodies and objects in normal instruction. Comparative experiments revealed that the body metaphor setting supported smoother collaboration than the conventional face-to-face metaphor setting. More detailed ethnomethodological analysis of the collaboration in the body metaphor setting following the comparative experiments also pointed to advantages of the ‘body metaphor’ concept. In particular when an instructor started and continued an instruction, the body metaphor setting provided important resources not only for the instructor but also for the operators to organize the situation of instruction. In the last section, remaining problems of the current ‘body metaphor’ implementation and authors' several current efforts for them are explained.
The simple recurrent network (SRN) proposed by Cleeremans & McClelland (1991) is one of the most successful models for implicit learning of complex sequences. In this study, we examined whether the SRN can also explain subjects' performance in explicit learning of an artificial grammar. In two experiments, subjects learned syntactic rules in the order that the SRN model predicted. This suggests that the underlying mechanisms of implicit and explicit sequence learning are closely related to each other in behaving like the SRN.
The purpose of the present study is to examine the nature of information that is available to the human sentence processing mechanism (i.e., the parser). We focus on the two types of information : “location information” and “grammatical information”. The experiment reported here was conducted in order to investigate which type of information is crucial for parsing two types of empty subject sentences: “subject-oriented” and “object-oriented” sentences. In a subject-oriented sentence such as “Taroo-ga Hanako-ni Tokyo iki-o hakuzyoosita (Taroo confessed to Hanako to go to Tokyo.)”, the person who will go to Tokyo is the subject “Taroo”. On the other hand, in an object-oriented sentence such as “Taroo-ga Hanako-ni Tokyo iki-o meireisita (Taroo ordered to Hanako to go to Tokyo)”, the object “Hanako” is supposed to go to Tokyo. Employing an on-line experimental technique (i.e., “dichotic listening” method), Oda et al. (1997) examined how the parser processes sentences such as the examples mentioned above. They found the “subject preference” phenomenon: the grammatical subject was preferred as the candidate for the empty subject of the embedded clause. However, no explanation was given for the reason of this outcome. Was the “preference” due to the “location” of the grammatical subject in the sentence initial position? Or, was the “preference” due to the grammatical subject bearing the grammatical function as the “subject”? In order to examine the word order effect, thus, we conducted an experiment employing the same procedure as that of Oda et al. except that the order of the subject and object in the sentence was exchanged. The results showed that, regardless of the change of the word order, there was the same “subject preference” effect that was observed in Oda et al. The findings of our experiment revealed that the surface order of the subject and object does not affect the tendency in which the grammatical subject is preferred as the candidate of the empty subject. This would suggest that the human parser does not make use of location information but rather makes use of grammatical information in sentence processing.