Calis paradigm with modifications by Jan Sterenborg
Description of the "Calis-Sterenborg paradigm"

In this study I used Calis double stimulation paradigm as described in the publication in Acta Psychologica (1984). Three modifications were made. In the presentation of the stimulus material: the separation of the first and the second stimulus-set and the time-relations were altered. Finally the hierarchy hypothesis was replaced by the process-hypothesis.
When using the hierarchy hypothesis one has to investigate two related concepts as whether in the 'process' hypothesis one has to investigate only one concept. By sex-concept is meant, that a person can discriminate between male and female persons.


A person who tries to identify another person, uses unconsciously a sex-concept as mediator.



Two female subjects participated in the experiment.


Training Phase

   The subjects were asked to identify four faces,  two male and two female faces. The faces were presented to them by slides. All faces were unknown to the subjects.
   In this phase the subjects were trained intensively in order to make them fully familiar with the faces on the slides. In this way the experimenter was always able to verify whether the subject had the right task interpretation.
   After the subjects were trained they were asked to perform in the actual identification task.

Task Phase

   In the task phase the question to be answered by the subject is still the same as in the training phase. However, the task is made more difficult in a sense that the researcher not only showed the familiar face, but preceding the familiar face he showed another face. This was not a face from the set faces used in the training phase, so in this way unfamiliar to the subject.


In our experiment we used stimulus-pairs in the following combinations:
   Four target faces were used; two male and two female faces. These target faces the subject had to identify and they appeared always as the second slide.
   The first stimulus of the pairs used, was picked from a set of sixteen faces; eight female and eight male faces. The faces used as first stimulus were different from those of the second stimulus!
   We presented five blocks of sixteen pairs. Within each block the stimuli were randomised without replacing. So in every block there were eight pairs of male-female and female-male combinations (Combdiff); and eight pairs of male-male and female-female combinations (Combsame).
Each sort of combination was presented in two presentation times; 50/30 msec. (Timrel1) and 70/10 msec. (Timrel2).
   The subjects could steer the experiment themselves by pushing a start-button after which the first stimulus pair appeared on the screen; after that they could make a choice by pushing one of four response-buttons which were related in the training-phase with the four target faces. After the choice was made the next stimulus-pair could be called by pushing the start-button and so on.
   When we combine the two time-relations with the two pair relations, we get four cells and within each cell four stimulus pairs (table 1.).

Table 1

Distribution of number of stimulus pairs in one block related to pair-combination and time-relation.

                         Timrel1   Timrel2
 Combsame        4             4     |    8
 Combdiff             4             4     |    8
                              8             8     |  16


   Two random-access-slide projectors with electronic programmable shutters were used in the experiment. The projectors, the shutters and the response panel were connected to a computer. The computer registered the stimulus-pair; the time-relation of the stimulus pair and the response given.


 When a person uses a sex-concept while identifying another person we expect: a significant relation between the two time relations and the number of correct identifications. The number of correct identifications will be lower in time relation Timrel2, because there is only 10 milliseconds to identify the second photo with respect to time relation Timrel1 where there is 30 milliseconds given to identify the second photo.
   Secondly, and this is crucial to our experiment. When the subject uses a sex-concept then, although performance becomes worse because of less time given, the probability of correct identification will be relatively better when the first and the second stimulus are in correspondence with each other. Correspondence means in this case that the two stimuli are identical with respect to the sex-concept. The stimulus pairs which are in correspondence with each other with respect to the sex-concept are: the male-male and female-female pairs (Combsame condition). Relatively better than when the first and second stimulus are not in correspondence with each other (Combdiff condition). This effect will be greater, the less time given to identify the second stimulus! So we expect a significant first order interaction between the two time relations and the two stimulus-pair combinations; male-male and female-female as compared to female-male and male-female pairs.


The results are presented in figure 1.

                     Subject    R                                    Subject     E

                        Timrel1       Timrel2                       Timrel1       Timrel2
Combsame      20                 13                               19                12
Combdiff          19                   4                                20                  4
      fig. 1. Results of subjects R and E in absolute number
                and Time-relation vs correct responses.

   As can be seen from a comparison of predictions and results there is enough evidence to say that our hypothesis is confirmed. This is for both subjects the case.
   Although both subjects performed (behaved) quite differently; the first one said nothing during the actual task phase; the other subject was constantly in doubt whether or not she had seen the correct person, the results over eighty trials are quite similar.
   Is the knowledge of interest within this person hierarchical organised? This question is now easily to investigate: you have to investigate the two concepts separately and when they are both active(this means a significant first order interaction for each concept) within this person you can establish whether the one concept is the necessary first step with respect to the second concept. In terms of statistical analysis this means a significant second order interaction.
                       Both pictures showing different persons
                       with the same sex.
                                     Both pictures showing different persons
                                     with different sex-identities.