In 1920, with the intention of measuring general intelligence without relying on language, Samuel Calmin Kohs of the University of Oregon introduced a new test in the Journal of Experimental Psychology, the block-design test. The test requires reproducing a drawing (a meaningless geometric design) using blocks. The sides of the blocks are either all one color (red, blue, white, or yellow) or divided into two colors on the diagonal (blue and yellow, or red and white). The test consists of 17 items with designs made up of 4, 9, and 16 blocks. The number of points earned on each item depends on the amount of time spent and the number of blocks correctly placed. Kohs blocks thus came to be.

Many tests based on the same principle have since been derived from the original test. The modifications have usually involved retaining two colors only, changing the design based on a variety of criteria, or simplifying the scoring mode. Block-design tests have been introduced into most major intelligence scales, including the Alexander scale in 1935, the Wechsler scales in 1939 (the blocks are part of the "performance" scale), and more recently in 1983, the Kaufman test battery (where they are part of the "simultaneous processes" scale). Several versions of the test have also been generated for clinical use. Some examples are the Goldstein and Scheerer test in 1941 and the Bonnardel test in 1971. Block tests are still widely used by psychologists. It is a new version of the test that Paulette Rozencwajg, Denis Corroyer, and Patrick Altman are offering here today, in honor of Samuel Kohs.

There are several reasons why this type of test has been so successful and has lasted so long. I can readily see three possible explanations, which differ in nature and in importance. Firstly, block tests are a kind of puzzle, so they seem more like a game than a school exercise or scholastic aptitude test. As such, they are stimulating for the subject. And despite the fact that timed execution in an evaluation setting is sometimes stressful, block tests are generally well-accepted, especially by children with academic lags. But the fun-filled aspect of the test cannot alone account for its longevity. What it assesses must also be of interest.

Secondly, the high correlations between Kohs blocks and composite intelligence tests prove that the block test does in fact assess a relatively general intelligence factor, a fact that was far from obvious at first. Factor analyses showing how block tests fit into the overall organization of cognitive abilities indicate that they load heavily onto the spatial visualization factor, which -- along with fluid intelligence, crystallized intelligence, long-term memory retrieval, and learning and memory -- is one of the main factors located just below the general factor in hierarchical models. The spatial visualization factor itself loads heavily onto the general factor, which is very close to fluid intelligence. Thus, Kohs blocks do not measure an isolated ability but a relatively general capacity. This is another useful feature.

The third reason is that behavior is easy to observe during block test execution. In most paper-and-pencil tests, subjects have to choose the correct answer among several choices, and very little time is allotted for responding. As a result, there are no behavioral clues to the subject's mental activity. With Kohs blocks (and in performance tests in general), the situation is totally different. The subject's mental activity is in some sense materialized by the choice of what blocks to use and where to place them. And because task execution takes place relatively slowly, there is enough time to observe the solving process. This is another one of the merits of this type of test. Clinical psychologists who use block tests never settle for simply noting a subject's performance level: they always want to see how the subject goes about accomplishing the task.

However, the observability feature is also a source of frustration for psychologists who use Kohs blocks. Because the block test was designed solely for measuring performance, no indexes for assessing the solving process are included. Testers hoping to study the process have attempted to define their own meaningful behavior sequences and indexes, in real time, without any predefined criteria or standards. This is an impossible task. Thus, despite the information-packed observation setting this test provides, it cannot be utilized in a satisfactory manner. The main advantage of Samuel is precisely that it characterizes subjects according to the methods they use. This is what makes this version of Kohs blocks radically different from its predecessors.

Samuel is based on Paulette Rozencwajg's research dealing with the strategies subjects use to solve Kohs blocks. In her studies, Paulette Rozencwajg of course detected the well-known analytic and global strategies. But she also found a strategy that had never been observed, the "synthetic" strategy. She defined each of these strategies in terms of precise behavioral criteria and then checked them for proper consistency before submitting them for inclusion in the software.

The system proposes nine indexes. Two are measures of performance: success/failure and solving time. The other seven, which capture the subject's activity and the solving process, are the number of times the subject looks at the design, total design inspection time, mean design inspection time, ability to correct mistakes (segmentation index), number of tries needed to put a block in place (orientation index), placement by rows and columns, and placement by gestalts. An observer equipped with only a pencil and a timer, no matter how attentive and competent he or she might be, obviously could not write down all the information needed to compute these indexes, let alone process them. So Samuel is a computer system. But what exactly does computerization have to offer? Unlike most available computerized tests, which merely change the testing medium, Samuel is not just a straightforward computerized rendition of the initial hand versions. In such cases, computerization is useful, granted, because it eliminates tedious tasks and provides up-to-date standardizations. But fundamentally, nothing is really different: subjects are still characterized solely in terms of their performance. In Samuel, computerization fulfills a totally different function -- it provides the tester with indexes that describe subjects in terms of the way they proceed in solving the problem. This feature makes Samuel one of the rare examples of how the computer can be used to make an old test into a truly new one, one that supplies information the old test could not furnish.

When a subject uses the analytic strategy, there is proper segmentation of the design, good orientation, placement by rows and columns, and frequent looking at the design. The synthetic strategy (the most efficient one) also involves good segmentation and good orientation, but the blocks are placed by gestalts and the subject does not look at the design as often. In the global strategy, where the subject proceeds by trial and error, both segmentation and orientation are inadequate, and the blocks are not placed by rows and columns, nor by gestalts. Samuel tells us which of these strategies is the closest to the subject's, i.e., the one the subject used the most. This initial diagnosis, which assigns a strategy to each subject, can then be further refined by taking the various behavior indexes into account. For instance, one can compare the costs of the different strategies by examining the total solving time, the total inspection time, and the mean inspection time, three indexes which do not enter into the strategy definitions. The results of a given item can also be examined, because the system outputs separate behavior indexes and strategies for each item. This feature is useful for determining whether the subjects are consistent in choosing a strategy.

Most tests currently in use are old ones, and as a result, they bear the mark of the state of the art in psychology and the objectives that prevailed at the time they were designed. In 1920 and even much later, mental functioning was seen almost exclusively in terms of efficiency, and very little energy was spent analyzing the psychological processes through which efficient mental functioning was achieved. Tests were mainly used for diagnostic purposes. For school children, for example, the goal was to predict academic success ... or failure, and testing proved helpful to counselors for providing guidance to students. The initial version of Kohs blocks is clearly representative of this period. But research in cognitive psychology has since shed new light on a number of aspects of mental functioning, and as psychology has changed, so have our requirements for such tests. Today's tests must supply useful information that can promote learning and teaching, something which conventional tests, with their focus on performance, simply do not do. For a good twenty years now, we have been hearing about how cognitive psychology has made many contributions that can be put to fruitful use in the design and development of evaluation procedures suited to this new learning-oriented goal. But actual applications are few and far between. Thus, we are very happy today to be able to provide psychologists with a test like Samuel. We now have a test that fits right into modern-day psychology and meets the new demands of practitioners.

Michel Huteau, Professor

INETOP