THE EMOTIONS AND THE DIMENSIONS OF DISCRIMINATION AMONG THEM
IN DAILY LIFE

by
Ilan Shalif

RESULTS II

(=====Back to result I.
To the discussion ---->
-48-

The subjects' dimensional scores for the ten dimensional analysis
The subjects' dimensional scores were computed according to the procedure that appears in the chapter on methods, which can be summarized as follows: For each subject, the row of his 48 scores of the unmanipulated 48 facial expressions was multiplied by the matrix of the dimensional coordinates of the ten dimensional S.S.A.-I analysis of those 48 items. (In that matrix there are 48 rows and 10 columns.) The result of this multiplication - for each subject - is a row of 10 scores - one score for each dimension.

The matrix of correlation of the ten dimensional scores

Among the 45 (non redundant) correlations of that matrix, none is significant. (The highest, is of r=-0.1167, p<0.9, two-tail.) This result indicates that the contents represented by the dimensions of the mathematical analysis - the dimensions of discrimination between the emotions of daily life are relatively independent.

The correlations between the 48 items and the ten dimensional scores

These correlations can determine the measure in which each dimension represents a variable that is common to many items or a content that is unique to one or two items. They may help to identify items that are not of the same domain of that the majority of the items are.

Results show that the significance (two-tail) of 30 of the 48 correlations with the first dimension is p<.002 (.002 These results and the two dimensional projection maps of the items (for each dimension against each of the others) reveal a relatively homogeneous dispersion of the items in the ten dimensional S.S.A.-I analysis. The combination of the two kinds of finding, indicates that the main content of all the 48 items and all the ten dimensions are of the same emotional domain.

-49-

The correlations between the dimensions and the items of basic emotion The correlations between these items and the ten dimensions are the most important means for the objective and non verbal interpretation of the content of dimensions of discrimination among the emotions of daily life. These items were scored twice - once on the first task, each on its own, and the second time - on the third task, by Q-Sort in which they were divided into 4 groups.

For the first task - the "free grading": 8 items' scores have correlations with the first dimensions of the S.S.A.-I analysis that reach the significance of p<.002, two-tail (the higher r=0.60) additional 9 have the significance of .002 As a whole, 57 of the 330 correlations are of p<.02 (less than 7 are expected by chance alone); 31 of these are of p<.002 (when by chance only 0.7 - less then one such high correlation is expected).

For the Q-Sort procedure 11 items' scores have correlations with the first dimensions that reach the significance of p<.002, two-tail (the higher r=0.53) additional 7 have the significance of p>.02; for the second dimension only one reaches the significance of p<.02; for the third - one is high and 4 are low; for the fourth there are only 4 of the lower significance; for the fifth only a low one; for the sixth, seventh, eighth and ninth dimensions there is no correlation of p<.02; the tenth dimension has three correlations with significance of p<.02 each.

As a whole, 32 of the 330 correlations are of p<.02 (when by chance less then 7 are expected); 12 of these are of p<.002 (less than one is expected by chance).

-50-

From the above correlations it can be concluded that the Q-Sort procedure is not a simple retest of the first task. It can also be concluded that the ten dimensions include in them contents that are not covered by the 7 basic emotions of the main list of this study or by the additional contempt, shame and distress items.

The correlations between the dimensions and the 24 artificial mixtures of basic emotion

The correlations between those items (materials: 1c) and the ten dimensions are very similar to those of the 33 items of basic emotions. For the free gradings 36 are of p<.02 and 21 of them are of the p<.002 level. For the Q-Sort - 20 are of p<.02 and 8 of them are of the p<.002 level. These mixtures were originally intended to be of a narrower spectrum of emotions than the 33 items of the basic emotions (materials: 1b). However the distribution of their correlations with the ten dimensions are very similar to that of the 33 items of 1/b.

The correlations between the 10 dimensions and the 148 words

126 of the words of materials: 2 (the second task) are correlated to the subjects' scores for the ten dimensions of the unmanipulated facial expressions of emotion (mixed expressions of daily life) - p<.02, two-tail (only 3 are expected by chance). These correlations are divided unevenly) among each of the ten dimensions. 89 words are significantly correlated to the first dimension - p<.02, two-tail (65 of them are of p<.002). For the second dimension 23 are of p<.02 and 27 are of p<.002). For the third there are 10 of the p<.02; for the fourth 10 of p<.02 and one of p<.002; for the fifth 14 of p<.02 and 4 of p<.002; for the sixth - 4 of p<.02; for the seventh - 3 of p<0.2 (as is expected by chance alone); for the eighth 4 are of p<0.02 and one of p<.002; for the ninth, 6 are of p<.02 and one of p<.002; for the tenth dimension there are only 4 correlations of p<.02 (two-tail).

-51-

Those correlations seem to stress three main points: first, the words as used in this study are very significantly related to the concrete emotional contents of daily life; secondly, the concrete emotional contents of the words are mainly related to the first five of the concrete emotional dimensions; thirdly, the difference between the distribution of the correlations of the words and the distribution of those of the artificial basic emotions and mixtures of them, indicates that the two kinds of items (or at least the two kinds of sampling of this study) sample the emotional domain in different ways. The narrower dispersal of the words is more noteworthy as the words were intended to be of an unrestricted wide spectrum while the above artificial- facial- expressions' items were of a restricted list of contents.

The correlations between the dimension scores and those of the 9 subgroups of basic emotions

The correlations between the subgroup scores and the dimension scores are the basic means for interpreting the relations between the basic emotion brain structures and the dimensions of discrimination among the emotions of daily life. Though the content validity of the items and the subgroups was found to be wanting, they are generally a sound enough base for the interpretation of the contents of the dimensions and for the examination of the hypothesis of this study.

In the following Tables 9 and 10, there appear the correlations between the dimensions of the mathematical solution of the S.S.A.-I analysis of the 48 unmanipulated facial expressions of emotions (materials: 1a of the first task of free grading) and the subgroup scores of the basic emotions (materials: 1b). Table 9 is for the free-grading (task 1) of the items of the 9 subgroups and Table 10 is for the Q-Sort procedure of those items.

-52-
Table No. 9: The correlations between the 9 subgroups of basic emotions  of
the free grading (task 1) and the dimensions of  the  S.S.A.-I  analysis of
the 48 unmanipulated facial expressions of mixed emotions (of task 1)
   ------------------------------------------------------------------------
   |Dimension|  1  |  2  |  3  |  4  |  5  |  6  |  7  |  8  |  9  |  10  |
   |Emotion\ |     |     |     |     |     |     |     |     |     |      |
   |---------|-----|-----|-----|-----|-----|-----|-----|-----|-----|------|
   |Happiness|0.63!|0.13 |-.05 |0.10 |0.07 |-.07 |0.16+|0.03 |0.12 |-.03  |
   |---------|-----|-----|-----|-----|-----|-----|-----|-----|-----|------|
   |Surprise |0.25!|0.09 |-.11 |0.06 |0.17*|-.03 |0.06 |-.16+|-.02 |0.02  |
   |---------|-----|-----|-----|-----|-----|-----|-----|-----|-----|------|
   |Interest |0.16*|0.24!|-.07 |0.01 |0.14+|-.02 |0.02 |0.07 |0.07 |0.04  |
   |---------|-----|-----|-----|-----|-----|-----|-----|-----|-----|------|
   |Fear     |-.22!|-.05 |-.30!|0.03 |0.11 |-.00 |0.10 |-.08 |-.16+|-.05  |
   |---------|-----|-----|-----|-----|-----|-----|-----|-----|-----|------|
   |Anger    |-.27!|-.01 |-.25!|0.13 |0.11 |-.03 |0.01 |-.11 |-.16*|-.04  |
   |---------|-----|-----|-----|-----|-----|-----|-----|-----|-----|------|
   |Disgust  |-.14+|-.00 |-.29!|0.04 |0.12 |0.06 |0.05 |-.07 |-.15+|-.02  |
   |---------|-----|-----|-----|-----|-----|-----|-----|-----|-----|------|
   |Sadness  |-.35!|-.02 |-.24!|-.01 |0.13 |0.11 |-.01 |-.06 |-.13 |-.07  |
   |---------|-----|-----|-----|-----|-----|-----|-----|-----|-----|------|
   |Contempt |-.16+|0.22!|-.12 |-.00 |0.07 |-.03 |-.00 |-.06 |-.06 |-.10  |
   |---------|-----|-----|-----|-----|-----|-----|-----|-----|-----|------|
   |Shame    |-.27!|0.25!|0.02 |-.12 |0.04 |0.03 |-.15+|0.04 |-.05 |0.01  |
   |======================================================================*

   "!" p<.002;
   "*" .0021>p and p<.02; 
   "+" .02 >p and p<.05;
   all two-tail. r=0.116 p=.05 one tail

Table No. 10: The correlations between the 9 subgroups of basic emotions of
the Q-Sort (task 3) and the  dimensions of  the  S.S.A.-I  analysis  of the
48 unmanipulated facial expressions of mixed emotions (of task 1)
   ------------------------------------------------------------------------
   |Dimension|  1  |  2  |  3  |  4  |  5  |  6  |  7  |  8  |  9  |  10  |
   |Emotion\ |     |     |     |     |     |     |     |     |     |      |
   |---------|-----|-----|-----|-----|-----|-----|-----|-----|-----|------|
   |Happiness|0.56!|-.05 |0.04 |0.10 |-.03 |-.07 |0.16+|0.02 |0.17*|0.01  |
   |---------|-----|-----|-----|-----|-----|-----|-----|-----|-----|------|
   |Surprise |0.31!|-.02 |0.05 |0.05 |0.01 |-.08 |0.07 |-.10 |0.11 |0.05  |
   |---------|-----|-----|-----|-----|-----|-----|-----|-----|-----|------|
   |Interest |0.07 |-.02 |0.08 |-.01 |-.05 |-.01 |0.01 |-.04 |0.04 |0.20* |
   |---------|-----|-----|-----|-----|-----|-----|-----|-----|-----|------|
   |Fear     |-.15+|-.11 |-.19!|-.15+|-.04 |0.13 |0.04 |-.10 |-.07 |-.14+ |
   |---------|-----|-----|-----|-----|-----|-----|-----|-----|-----|------|
   |Anger    |-.23!|0.03 |-.11 |0.22!|0.07 |-.09 |0.01 |-.00 |-.01 |0.02  |
   |---------|-----|-----|-----|-----|-----|-----|-----|-----|-----|------|
   |Disgust  |0.03 |-.11 |-.08 |0.04 |0.05 |-.02 |-.03 |0.03 |-.09 |0.10  |
   |---------|-----|-----|-----|-----|-----|-----|-----|-----|-----|------|
   |Sadness  |-.42!|-.04 |-.04 |-.06 |0.01 |0.06 |-.08 |0.07 |-.13 |0.02  |
   |---------|-----|-----|-----|-----|-----|-----|-----|-----|-----|------|
   |Contempt |-.06 |0.17*|0.11 |-.16*|-.13 |0.01 |-.03 |0.04 |0.05 |-.08  |
   |---------|-----|-----|-----|-----|-----|-----|-----|-----|-----|------|
   |Shame    |-.33!|0.11 |0.14 |-.04 |-.10 |0.08 |-.17*|0.11 |-.05 |-.02  |
   |======================================================================
   "!" p<.002;
   "*" .0021>p and p<.02; 
   "+" .02 >p and p<.05;
   all two-tail. r=0.116 p=.05 one tail
-53-

The comparison of Table 9 with Table 10 reveals, once again, effects that are noted at the beginning of the chapter where subgroup validity was examined. It can be seen that surprise and interest of the first task were highly related to the dimensions of discrimination among emotion (three significant correlations) and poorly so in the Q-Sort task. For the interest subgroup of the Q-Sort procedure there is no dimension with which both items are correlated.

For the four negative emotions of the basic list - fear, anger, disgust, and sadness, there are only small differentiations in the first task. In Table 9 of that task, one can see that they have very similar correlations with the first, third, fifth and ninth dimensions.

In Table 10 of the Q-Sort task, one can see that the differentiation among the relatively more valid subgroups - fear, anger and sadness - is sharper. The correlation of the disgust subgroup which is less valid, becomes insignificant.

The two subgroups of contempt and shame that were included for control purposes are also more differentiated in the Q-Sort procedure. Shame has nearly the same pattern with the exception of the second dimension which is more clearly the "contempt dimension" on the Q-Sort.

Contempt itself, changes its pattern greatly on the first, third, fourth and fifth dimensions. Only its correlation with the second dimension is stable. These findings converge with the nonsignificant correlation between the scores of the two procedures. They contribute to the conclusion that the common denominator of the scores of the two procedures of this subgroup, - which keep it correlated to the second dimension - is not contempt.

The difference between the patterns of the correlations of the subgroup scores of the two procedures, and dimensional scores (of the 48 1a items) highlights the fact that the subjects changed the parameters of their discrimination-activity when the task changed.

-54-

At this level of analysis it is too early to assess the relative contribution of the various factors which cause this. It may be that two main factors were involved: The first is the change in the relative contribution of various parts of the face to the assessment. (As Ekman & Friesen, 1975, have shown - the willful control of the various parts of the face is not of the same level of success.) The second is a Phenomena which Hirschberg (1980) pointed out. She found in her study that subjects change the dimensions of discrimination as the task changed.

Another uncontrolled variable which contributed to the difference between the two procedures is subjects' attitudes: some subjects responded with strong emotional distress or relief to the change from free grading to Q-Sort. It is noticeable that there is only a small measure of exclusiveness with regard to the relations of the subgroups and the dimensions of the S.S.A.-I analysis of the 48 mixed expressions.

Ideally for the theory, each subgroup should be correlated to a dimension of its own. But due to the characteristics of the mathematical solutions and their restrictions this seldom happens. The most restricting obstacle is that till now, it is theoretically possible to achieve but hard to attain, systematic rotations of the axes (dimensions) of the mathematical multidimensional model. More problems stem from the relatively low content validity of some of the subgroups which also contributes its share to the blurring of the results. However, the integration of the results of the two procedures is enlightening and enables us to examine the hypothesis of this study.

The directions of the emotions in the space of the S.S.A.-I solution
When one is tracing the directions in the mathematical multidimensional model of the S.S.A.-I analysis, not in the usual way of visually scanning the scatter of the items but in the new adaptation of the S.S.A.P.-I, one is mainly relying on correlations. The correlations are those that are between subjects' dimensional scores and subjects' scores for the relevant items and variables included in the study but not in the S.S.A.-I analysis.

-55-

For instance, if one is examining the directions of happiness and sadness according to Table 10, it can be seen that scores of three dimensions are significantly related to the happiness or the sadness subgroups of basic emotions - namely the first, the seventh and the ninth. As a result of this examination it can be concluded that the direction of the two subgroups' content, in the ten dimensional space-model, can be traced as a bipolar vector which may be contained in the three dimensional (cube) sub-space of the first, the seventh and the ninth.
Plate No. 1: The direction of happiness and sadness in the plane of first and
ninth dimensions.
                              first dimension
              happiness - r=+0.56 ^   *H
                                  |   *a
                                  |  *p
                                  | *p
                                  | *i
                                  |*n
                                 S*e              happiness - r=+0.16
                 ----------------a*s--------------> ninth dimension
sadness - r=-0.13                d*s
                                n*|
                               e* |
                               s* |
                              s*  |
                sadness - r=-0.42 |

    dimension.
A two dimensional projection of that cube - the projection of the first dimension against the ninth dimension is shown in the above Plate. In it, the equivalent bipolar vector is the sum of the two one-dimensional bi-polar vectors which represent the subgroups' contradictory correlations with these two dimensions.

The directions of the subgroups which are delineated below are according to the pattern of the correlations of Tables 9 and 10 (with the help of the correlations of the items themselves when needed).

-56-

The directions of the happiness is mainly along the first dimension (which is mainly the happiness-sadness dimension) and to a lesser extent with the seventh and the ninth.

The directions of surprise is along the first, the fifth (which is mainly the surprise dimension) and the eighth dimensions and it looses its uniqueness in the Q-Sort task.

The directions of interest - which seems to be of very low content validity - is along the second dimension (the only dimension with which both groups' items are correlated - and only in task 1). The high significance correlation of this subgroup-score of the Q-Sort, with the tenth dimension is due to item 24 which seems to be loaded with the content of disgust.

The directions of fear is most prominent on the third dimension (the fear dimension) and with a lesser measure along the first dimension (opposite that of happiness) and the fourth dimension (opposite that of anger - as is often found in other studies).

The directions of anger is along the first dimension (opposite to happiness and along the fourth dimension (which is the anger dimension).

The directions of disgust which has a low content validity is not based on the subgroup score. It is based on the items which were found to be highly loaded with the content of disgust. Its direction is mainly along the tenth dimension (the disgust dimension).

The directions of sadness is mainly along the first dimension - in the opposite direction of happiness. Though it has a very clear direction (second only to happiness) it does not have an independent direction of its own.

The directions of contempt is along the second dimension - a direction of its own (the pro-social dimension of interest and compassion) and along the fourth dimension - in the opposite direction of anger, which is not logical if the content of this subgroup is really contemptuous.

-57-

The direction of shame is along the first dimension - in the direction of sadness and in the opposite direction of happiness and along the seventh dimension (the shame dimension) - here too in the opposite direction of the happiness subgroup.

Summing up, the above results give a substantial support for hypothesis one. It was found that all seven basic emotions of the study (and the two additional ones) are related to the dimensions used by the subjects when they discriminated between the emotions of daily life. The main contradiction between the results and the theory of this study is that sadness is not an independent variable but an opposite one to the happiness.

The differences between the directions of the 9 subgroups in the model
The best measure of the relative independence between the subgroups would be the computation of the angles between their directions in a model which is beyond our resources. The second best is the testing of the significance of the differences between the correlations of the subgroups and the dimensions of the S.S.A.-I analysis.

The following table 10 contains the essentials of this examination. The upper right- hand triangle is of the first task free gradings and the lower left-hand triangle is of the Q-Sort procedure of the third task. In it, one can see that all pairs of subgroups differ significantly. But, we cannot be sure about the measure in which they are independent of each other when all ten correlations are considered simultaneously, as there is still not available a suitable test for this purpose.

*********************Table 11 about here*********************

The significance of the differences demonstrated in Table No.11

In an array of coefficients of correlations as in other multivaraete tests, the first step is the test for all the results simultaneously. In the matrix of correlations between the 9 subgroups of basic emotions and the 10 dimensions of the 48 items of mixed and unmanipulated expressions of each procedure there are 90 coefficients.

-58-
Table No. 11: The difference between subgroups' correlations in task 1 and  3
with the dimensions (and pairs of them) of the 48 items of materials: 1/a  of
task 1*. The higher triangle is for task 1, the lower is for task 3.
_____________________________________________________________________________
Sub  *Happi  *Sur    *Inte   * Fear  * Anger *Disgust*Sadness*Con    * Shame
group*   ness*  prise*   rest*       *       *       *       *  tempt*
 Ha  *  D1;D7* D1=.38* D1=.47* D1=.85* D1=.91* D1=.78* D1=.98* D1=.79* D1=.91
 pp  *  *    * 0.0001* 0.0001* 0.0001* 0.0001* 0.0001* 0.0001* 0.0001* 0.0001
 in  * D1 *  *D5+8=17*D2+5=13*D3+9=44*D3+9=41*D3+9=43* D3=.19* D2=.10* D7=.31
 ess * D7;D9 * 0.0047* 0.0384* 0.0001* 0.0001* 0.0001* 0.0287* 0.0668*  0.001
-----------------------------------------------------------------------------
 Su  * D1=.25*  D1;D5* D8=.24* D1=.47* D1=.53* D1=.40* D1=.60* D1=.41* D1=.53
 rp  * 0.0001*  *  D8* 0.0003* 0.0001* 0.0001* 0.0001* 0.0001* 0.0001* 0.0001
 ri  *D5+8=13*    *  * D2=.14*D3+9=23*D3+9=23*D3+9=22*       * D2=.13*D7+8=32
 se  * 0.0516*D1     * 0.0192* 0.0005* 0.0047* 0.0013*       * 0.0446* 0.0002
-----------------------------------------------------------------------------
 I   * D1=.50* D1=.25*  D1;D2* D3=.27* D4=.22*       * D1=.49* D2=.20* D1=.39
 n   * 0.0001* 0.0026* *   D5* 0.0048* 0.0132* see   * 0.0001* 0.0301* 0.0001
 t   *D10=.19*D10=.15*     * *D10=.34*D10=.18* discu *D10=.18*D2+3=22* D7=.17
 e   * 0.0212* 0.0526*see dis* 0.0005* 0.0392*  sion * 0.0392* 0.0179* 0.0495
 re  *       *       *cussion*       *       *       *       *D10=.28*D10=.22
 st  *       *       *    D10*       *       *       *       * 0.0030* 0.0188
-----------------------------------------------------------------------------
 F   * D1=.71* D3=.24* D1=.38*  D1;D3* D4=.10*  ***  * D1=.13* D2=.27*D2+7=41
 e   * 0.0001* 0.0080* 0.0001*  *  D9* 0.0069*       * 0.0087* 0.0001* 0.0001
 a   * D3=.23*       * D2=.28*    *  *       *       *       *D3+9=20*D3+9=29
 r   * 0.0179*       * 0.0003*       *       *  ***  *       * 0.0007* 0.0001
     *4+10=31*       *D3+9=31*D1;D3  *       *       *       *       *
     * 0.0021*       * 0.0001*D4;D10 *       *       *       *       *
-----------------------------------------------------------------------------
 A   * D1=.79* D4=.17* D1=.44* D4=.36*  D1;D3* D1=.13*  ***  * D2=.24*D2+7=28
 n   * 0.0001* 0.0548* 0.0001* 0.0001*     D9* 0.0049*   *   * 0.0001* 0.0002
 g   *       *       * D2=.25*       *  *    *       *   *   * D3=.12*D3+9=26
 e   *       *       * 0.0019*       *    *  *       *   *   * 0.0418* 0.0006
 r   *       *       *D3+9=27*       *       *       *   *   *       *
     *       *       * 0.0006*       *D1;D4  *       *  ***  *       *
-----------------------------------------------------------------------------
 D   * D1=.54* D1=.29* D1=.31*D10=.24* D4=.18*  D1;D3* D1=.21* D2=.23* D1=.13
 i   * 0.0001* 0.0026* 0.0001* 0.0129* 0.0446*  *  D9* 0.0001* 0.0002* 0.0485
 s   *       *(D9=.21* D2=.24*D10+D4=*       *    *  *       *D3+9=19* D2=.26
 g   *       *0.0294)* 0.0019*   0.31*       * see   *       * 0.0016* 0.0005
 u   *       *       *D3+9=30* 0.0018*       * discus*       *       *D3+9=28
 st  *       *       * 0.0001*       *       *   sion*       *       * 0.0001
-----------------------------------------------------------------------------
 S   * D1=.98* D1=.73* D1=.51* D1=.27* D1=.19* D1=.44*  D1;D3* D1=.19*D2+7=31
 a   * 0.0001* 0.0001* 0.0001* 0.0009* 0.0179* 0.0001*       * 0.0014* 0.0001
 d   * D9=.30*       * D2=.25*D3+4=16* D4=.28*       * *     * D2=.24* D3=.26
 n   * 0.0069*       * 0.0013* 0.0401* 0.0024*       *    *  * 0.0002* 0.0003
 e   *       *       * D3=.17*       *       *       *       *D3+9=14*
 ss  *       *       * 0.0256*       *       *       *D1     * 0.0212*
-----------------------------------------------------------------------------
 C   * D1=.62* D1=.37* D1=.32* D2=.28* D4=.38* D2=.28* D1=.36*  D1;D2* D1=.12
 o   * 0.0001* 0.0003* 0.0001* 0.0044* 0.0001* 0.0016* 0.0001*       * 0.0427
 n   * D2=.23* D2=.19*       * D3=.30* D2=.14* D4=.20* D2=.21*  *    * D7=.14
 t   * 0.0207* 0.0446*       * 0.0024* 0.0823* 0.0188* 0.0192*    *  * 0.0217
 em  *D2+4=35*D2+4=29*       *D2+3=43*D2+4=45*D2+4=36*D2+4=18*       *D1+7=15
 pt  * 0.0006* 0.0052*       * 0.0001* 0.0001* 0.0001* 0.0322*D2;D4  * 0.0113
-----------------------------------------------------------------------------
 S   * D1=.89* D1=.64* D1=.44* D3=.33* D4=.25* D1=.35*(D2=.15* D1=.27* D1;D2
 h   * 0.0001* 0.0001* 0.0001* 0.0014* 0.0045* 0.0002* p<.05)* 0.0004*     D7
 a   * D7=.33* D7=.24* D5=.11* D7=.21* D7=.18* D7=.14*(D3=.18* D7=.14* *
 m   * 0.0021* 0.0202* 0.1020* 0.0287* 0.0307* 0.0985* p<.05)* 0.0427*     *
 e   *D7+9=41*       * D7=.16*D3+7=41*       *D2+3=31*(nonsig* D4=.13*
     * 0.0002*       * 0.0244* 0.0001*       * 0.0013* corr.)* 0.0643*D1;D7
-----------------------------------------------------------------------------
* According to the Hotelling test for the difference between coefficients of correlations that are correlated (from Guilford, 1965, p190-191) as the majority of the subgroups' scores are correlated (see Table 2).
**For each difference one line for the size and one for the significance.

-59-

In the first task of free grading, 23 of them are of the two-tail significance of .05 (or higher), 4.5 are expected by chance alone. The significance of this finding is of p<0.00001. In the third task of Q-Sort there are 16 significant correlations and the significance of these findings is of p<0.00032.

The significance of the differences between the correlations of the subjects was assessed by the Hotelling test for correlated coefficients of correlations because the majority of the subgroups' scores are correlated (see Table 2).

Even when two correlations are significantly different mathematically, there is still a possibility that the subgroups are in the same direction of the multidimensional space of the mathematical solution of the S.S.A.-I. This can happen when the two subgroups have the same multidimensional direction (and emotional content) and differ only in the loading of the emotional content of the items due to technical problems. Due of this and the lack of a suitable test for the simultaneous testing of the ten dimensional differences between the correlations of each pair of subgroups - the following findings are a tentative first approximation.

The tests for the significance of the differences between subgroups were conducted only when at least one of the subgroups was significantly correlated with the subjects' score for that dimension.

In the free grading task there are 144 one-dimensional differences (out of 360) where at least one subgroup is significantly correlated to that dimension, as well as 152 two-dimensional differences (out of 1620) in which each of the dimensions of the pair is correlated to at least one of a pair of subgroups.

47 of the above 144 one dimensional differences are significant (combined probability of p<.0...(twenty zeros)015). Among the 152 two-dimensional differences, there were tested only those 98 in which none of the one-dimensional differences (between the two subgroups) reached p<001 in any of the relevant two dimensions (as in those cases the contribution of the two dimensional difference can not be of high value to us). 19 of the above 98 differences are significant and their combined probability is p<.0000165.

-60-

Table 11 - the upper right-hand triangle - shows the 47 significant one dimensional differences and the 19 two-dimensional ones. Only two of the 36 pairs of subgroups do not have even one significant difference. There are no significant differences (in this task) between the correlations (and the directions of the vectors that represent the subgroups in the multidimensional model) between fear and disgust and between those of anger and sadness.

In the third task of the Q-Sort there are 16 significant correlations between subgroups and dimensions (out of the 90 of the matrix) with combined significance of p<0.00032). 108 out of the 360 one-dimensional pairs of subgroups' dimensional correlations include in them one of the above 16 significant correlation.

42 of those 108 are significantly different (the combined probability of those differences is p<.0.....[twenty zeros]039). In this procedure too, there were tested two-dimensional differences among pairs of groups. Part of these as well as the 42 one-dimensional differences are on the left-hand lower triangle of Table 11.

One can see from the convergence of the results of the two procedures that the nine subgroups differed significantly from each other. One can see that all 36 pairs of subgroups have at least one significant difference. These results partially support the hypothesis of this study dealing with the differences between the directions of the basic emotions within the universe of discrimination among emotions.

However, these results also contradict our claim that sadness is a basic emotion on its own and as such relatively independent of the others. The findings about the bipolarity of happiness-sadness highlight the bipolarity aspects of the basic emotions - an aspect that was stressed by Darwin (1872), yet neglected by most of his followers.

-61-

The content of the 10 dimensions of the S.S.A.-I analysis, of the 48 facial expressions of unmanipulated emotional mixtures

According to our theory and hypothesis, the structure of the emotional domain, with regard to dimensions of discrimination among daily facial expressions and experience of emotion, is of nearly orthogonal dimensions. In this space the content of the dimensions is supposed to converge with that of the already-found basic emotions.

In the mathematical solution of the S.S.A.-I, the contents of the above dimensions are expected to converge with the axis-dimensions of the analysis. It is also expected - as was indeed found - that subjects' scores (which are computed with a simplified version of the S.S.A.P.-I) for these axes-dimensions will be relatively independent of each other.

In order to interpret the contents of the dimensions as objectively as possible in this study, we examined the correlations between these dimensions' scores, and all original item scores (non-verbal materials 1b and 1c of both procedures and the 148 words of task 2) as well as subgroup scores and the dimensional scores of the subjects. All of them are Spearman nonparametric correlations as the distributions of most of the scores are not "normal". All levels of significance are two-tailed unless it is specifically stated that they are one-tailed.

The first dimension - Happiness (pleasure) versus Sadness

The subjects' scores of this dimension correlate with those of the subgroup happiness: +0.63, and that of sadness: -0.42 - both p<.001. Its highest correlations are with the following words: happiness (+0.52); satisfied (+0.52); depression (-0.45); distress (-0.45) - all p<.001.

--62-

The second dimension - Interest in others + leniency versus drowsiness In the free grading procedure (as seen in Table 9) the contempt, interest and shame subgroups are correlated with this dimension - the only dimension with which both items of interest are correlated. In the Q-Sort procedure only the correlation of the contempt subgroup is still significant.

The highest correlations of this dimension with words - whose correlation with this dimension is higher than with the other dimensions (and positive) are with activism, attraction, concern, generosity, initiative, interest, leniency, patience, potency, respect, right, sincere, tenderness, wakefulness.
The highest are attraction (+0.30); concern (+0.30); and wakefulness (+0.32).
The words that have significant and negative correlation with this dimension are drowsiness, doze, fatigue.

It seems that the common denominator of the content of the items of the subgroup of contempt (as regarded by the subjects of this study and which is the cause of its correlation with the second dimension) contains leniency and potency components but not scorn. (See the section about the validity of the subgroups of basic emotions p37 and 42.)

The third dimension - the Fear dimension

In the free grading procedure all four negative emotions: fear, anger, disgust and sadness; are correlated with the third dimension (p<.002) but only for the fear subgroup all four items are significantly correlated with this dimension.
In the Q-Sort only the fear subgroup is correlated significantly with it (-0.19, p<.002).

The words that have significant and negative correlation with this dimension (in the same direction of fear) are: astonishment (-0.17); disgust (-0.20); droopy (-0.21); guilt (-0.16); panic (-0.21); remorse (-0.20); servility (-0.19).

The fourth dimension - the Anger dimension or firmness versus embarrassment

In the free grading, the higher correlation with this dimension is of the anger subgroup: r=0.13 0.05
-63-

Its higher correlations with words are: attraction (+0.21); courage (+0.18); decisiveness (+0.19); embarrassment (-0.20); hope (+0.18); stubbornness (+0.16); superiority (+0.19); weakness (-0.21) and yearning (+0.17) - all between p<.0021 and p<.02. The highest correlation is with firmness (+0.22, p<.002).

> The fifth dimension - the Surprise dimension (stupefaction)

In the free grading the surprise subgroup is correlated with this dimension (+0.17) as well as the interest one (+0.14); both are of p<.02. Those correlations are mainly due to item surprise no. 18 (+0.21) and for item interest no. 8 (+0.20) that was found in the examination of validity of items to be loaded with surprise rather than with interest. One item - disgust no. 5 is negatively correlated to this dimension (-0.22, p<.002).

In the Q-Sort procedure no subgroup and only two items are positively correlated with this dimension - one is of dubious content: disgust no. 15 and the other is of an artificial mixtures whose content is surprise + questioning (+0.20, p<02). One item - contempt no.7 is negatively correlated to this dimension.

The highest positive correlations with words are with anger, caution, compassion, craving, curiosity, daring, disappointment, doze, excitement, potency, superiority, suspicion, firmness - all of .002>p<.02 - and with appetite, skepticism and stupefaction - of p<.002. Not even one of the words is negatively correlated with this dimension. The most suitable words to summarize the above content is vigilance or wariness.

The sixth dimension - troubled versus lightness

None of the subgroups nor the items of basic emotions of the free grading procedures are correlated with this dimension. Only mixture no. 1 of "neutral" content is correlated here. In the Q-Sort we find nearly the same situation. Only fear no. 3 and mixture no. 23 of sadness + disgust reach the level of p<.02.

-64-

The words that are correlated with this dimension are: amused (-0.17); appetite (-0.18); boredom (-0.15); frustration (+0.15); guilt (+0.16); humorous (-0.18); preoccupied (+0.19); relief (-0.15); sadness (+0.14); stupefaction (-0.14) - all of 0.009 < p <.05.

The content of this dimension is: troubled (preoccupied- guilt- worry- frustration) versus relief.

The seventh dimension - the Shame dimension

The seventh dimension is correlated to happiness - both procedures +0.16 - and with shame - on the free grading: -0.15 and on the Q-Sort: -0.17. The words that are significantly correlated with this dimension are: bitterness (-0.15); depression (-0.15); envy (-0.17); happiness (+0.18); indifference (-0.18); rigor (+0.16); satisfaction (+0.14) - the significance of all is of .01 The basic emotion shame that was included for control purposes has only two items. However, it was found to be of a unique content of its own. It was previously found to be as such only in a small number of studies - due to its small part in the systematic variance. It seems that its clear appearance in our study is due to the strength of the study procedure and the new methodology on which it is based.

The eighth dimension - seems to have the content of contempt
This dimension is correlated in the free grading procedure with the surprise subgroup (-0.16 p<.05); with anger no. 4 (-0.15, p<.05); anger no. 12 (which is highly loaded with disgust - -0.14 p<.05); disgust no. 15 (not valid - -0.16, p<.022); surprise no. 10 (-0.22, p<.002).

In the Q-Sort procedure, only surprise no.18 (-0.16, p<.022) and one of the "neutral" mixtures (+0.14, p<.05) are correlated with this dimension. Among the words, the significant correlations are with anger (-0.14, p<.05); contempt (-0.24, p<.001); desire (-0.16, p<.021); disqualification (-0.20, p<.004); grateful (-0.20, p<.004); greediness (-0.14, p<.05); humiliation (-0.15, p<.05); misery (-0.19, p<.007) mockery (-0.14, p< .05) superiority (-0.19, p<.007).

-65-

As only one "neutral" item is positively correlated with this dimension (the highest correlation for words is restraint - +0.13 p<.08), we can rely only on the items that are on the negative pole. Though their common denominator is not entirely clear it seems to converge with the content of contempt.

The ninth dimension - love versus hate

In the free grading the subgroups of anger, fear and disgust are negatively correlated to this dimension (p<.05 to p<.02). In the Q-Sort only the happiness subgroup is correlated with it (+0.17, p<.02).

Among the words, the correlations are with concern (+0.16); decisiveness (+0.14) disappointment (-0.16); hate (-0.15); humiliation (-0.15); interest (+0.15); love (+0.15); satisfaction (+0.15) - all of p<.05, and amused (+0.19); depression (-0.17); longing (+0.19); rigor (+0.20); vigilance (+0.18) - all of p<.02; and contempt(-0.22, p<.002).

The examination of the correlations of love and hate with the other dimensions (their correlations with this dimension are higher than any of the other seven advanced dimensions), and the correlations of this dimension expounded above, lead to the conclusion that the content of the contradiction - love versus hate - converges with the content of this dimension.

The tenth dimension - the Disgust dimension

In the free grading none of the subgroups and only one of the items of the basic emotions - that of contempt no.33 (-0.17, p<.02) - are correlated with this dimension.

-66-

In the Q-Sort, only the non valid interest subgroup is positively correlated to this dimension (+0.20, p<.004 - due to the correlation of item no. 24). The other items of the basic emotions that positively correlate with this dimension are: disgust no.5 (+0.17, p<.02); disgust no. 21 (+0.17, p<.02); and basic emotion no. 12 (the item its content was intended to be of anger but was found to be mainly of disgust - +0.20, p<.004). The only significant negative correlation is of the subgroup fear (-0.14, p<.05).

Among the words, the significant correlations are with activity (-0.17, p<.02); conciliated (-0.14, p<.05); excitement (-0.14, p<.05); rigor (-0.18, p<.02); righteousness (-.21, p<.004); satisfaction (-0.21, p<.004).
From the above correlations it seems that the content of this dimension is of disgust versus satisfaction.

In summation of the examination of the content of the dimensions of discrimination among facial expression

The analysis of the matrix of correlations between the subjects' scores for the ten dimensions and all the other scores produces a relatively clear picture. This is so although the content validity of some subgroups, items and all the methodology of item building is deficient. This was made possible by the new technique which can be used to extract meaningful variance by the convergence of a plethora of scores.

The conclusion from the examination of the content of the dimensions is that the main dimensions converge with the main basic emotions as was predicted by the first hypothesis derived from the evolutionary theory of this study. However, there are some reservations:
The first reservation relates to the bi-polarity of the basic emotions: a) it was found that sadness is not an independent basic emotion; b) most of the 10 dimensions reveal a bi-polar content.

-67-

The second reservation is about the basic emotions interest and contempt whose content validity is questionable. The two items of interest are not heavily loaded with the same content, so that interpretation of the emotional content of interest had to be based on circumstantial evidence. The problems of the contempt subgroup and emotion are more complicated. The subgroup items have a common denominator, however, the subjects did not perceive the negative derogatory aspect or connotation of the items of this emotion.

Following the proposition of reservations, we conclude that the results essentially support the first concrete hypothesis of the study and at the same time supply the most needed for support for the family of evolutionary theories of emotion.

The dimensions of discrimination among the 96 words of emotion

In order to examine the feasibility of the second hypothesis of this study, the 96 words of emotion of materials 2a were analyzed by S.S.A.-I of Guttman-Lingoes series (Lingoes, 1973) - all the nine analysis of 2 to 10 dimensions were done. Then the subjects' dimensional scores were computed for the ten-dimensional analyses, by the Shalif et. al. (1981) adaptation of the S.S.A.P.-I of the same Guttman lingoes series. Next, the Spearman correlations between those dimensional scores and all the original scores of the subjects and their subgroup scores were computed.

The coefficient of alienation of the nine S.S.A.-I dimensional analysis is as follows: for two-dimensions - 0.169; for three - 0.121; for four 0.095; for five - 0.083; for six - 0.073; for seven - 0.063; for eight - 0.057; for nine - 0.051; and for ten dimensions - 0.047.

The two-dimensional analysis

The best way to represent the content of the dimensions of this analysis is by the words that are near the two margins of each dimension.

-68-

Those of the first dimension are: depression + restlessness which are near the margin of one side versus calmness + happiness + confidence which are near the margin of the other side.

Those of the second dimension are alertness + firmness on one side versus habitualness + serenity on the other.

The above two dimensions are of a clear concrete content. The first is of restless- depression versus its opposition - calm-happiness. It is clear how some theorists make of it abstraction and call it the "Evaluative dimension". The content of the second is of a firm alert (caused many times a day by the many challenges that occur in a competitive society), versus the habitual relaxation (of the background time between challenges of the less nervous people). It is clear how it can be "abstracted" to fit the "Dynamism- dimension" that was found frequently in two-dimensional analysis (according to articles by Snider & Osgood, 1969).

The three-dimensional analyses

Those near the margins of the first dimension are depression + distress + misery + frustration versus relaxed + at ease + satisfied + confidence. Those of the second dimension are alertness + firmness + hope + yearning versus habitualness + serenity + indifference + boredom. Those near the margins of the third dimension are pride + haughtiness versus fatigue + longing + caution.

The content of the first dimension is mainly of the concrete content of depression and distress versus its opposite - contentment. It can be seen how it matches that of the activity of the inborn structure of "separation distress" (or as it is called by Izard, 1971 - the basic emotion distress). The content of this dimension can be "abstracted" to fit the first dimension of the "thrifty" cognitive approach or that of the evaluative dimension of Osgood's semantic differential technique.

-69-

The content of the second dimension fits that of the developmental concept of wariness or its other version vigilance. In a way, it fits the internal activity of monitoring each ongoing intentional activity that meets the ongoing challenges of life. It is clear why those of the thrifty cognitive approach abstracted it to represent the Activity content. (It is also clear how it became Osgood's second dimension.)

The content of the third dimension fits that of the inborn structure that causes the bi-polar continuum of pride-shame which is better known by the name of the basic emotion shame. It was suitable to abstraction as the Potency dimension - the third dimension of the thrifty cognitive approach (and Osgood as well).

The above results are a clear-cut refutation of the second hypotheses of this study. They deal a severe blow to the theoretical efforts we made with the aim of reconciling the contradiction between the evolutionary theories of emotions and those of the thrifty cognitive approach.

In the following chapters of discussion and conclusion we offer a solution with the aim of reconciling between our results and our theory.

The ten dimensional analysis of the 96 words of emotion

In order to deepen the evaluation of the hypothesis of the study, the matrix of correlations between the subjects' dimensional scores for these ten dimensions and their 300 relevant scores was re-examined. (The 300 scores are of the 10 dimensions themselves; the 148 words of materials: 2a, 2b, 2c; of the 114 scores of the 57 artificial facial expressions of the two procedures - materials: 1b and 1c and the 18 subgroup scores of the 9 basic emotions of the two procedures; and the 10 dimensional scores of the analysis of the 48 mixed and unmanipulated facial expressions.)

The 10 dimensions' inter correlations show that only one of the correlations is significant - r=-0.20, p<0.004 (two-tail). This may be the result of uneven sampling of the emotional domain due to our unintentional bias or due to the bias of the language itself.

-70-

The significant correlations between the 9 basic emotions and the ten dimensions were spread unevenly. Most of the significant correlations are with the first four dimensions. The eighth and ninth dimensions do not have any significant correlation with any subgroup.

The following interpretation of the content of each of the ten dimensions of the 96 emotional words is mainly based on the convergence of information drawn from two sources: One is the content of the words that are on the two margins of each dimension - the classical way - and the other is the above 10 by 300 correlation matrix.

The first dimension - contentment versus distress

The words that are near the margins of this dimension are misery + distress on one side and calmness + at-ease + contentment on the other. Its highest correlation with subgroups are with happiness (+0.57) and sadness (-0.40). The words that have the highest correlations with it are: at-ease (+0.69); calmness (+0.74) contentment (+0.69); depression (-0.67); distress (-0.69); happiness (+0.67); sadness (-0.67); satisfied (+0.72) - p<.001 for all of them. The highest correlation between this dimension and any dimension of the 48 unmanipulated facial expressions of mixtures of emotions is with that of the happiness- sadness one - r=+0.60 p<.001. The content of this dimension is mainly of calmness, satisfaction and contentment versus depression, distress and sadness.

The second dimension - alertness versus serenity

The words that are on the edges of this dimension are alertness + tension + curiosity on one side and serenity + disrespect + lightness on the other. Its highest correlations with subgroups are with happiness (negative) and with sadness and shame (positive).

-71-

The words that have the highest correlation with it are: alertness (+0.47); complexity (+0.46) disappointment (+0.42); disregard (-0.27); happiness (-0.33); lightness (-0.26); satisfied (-0.30); serenity (-0.38); servility (-0.20); stubbornness (+0.40) and tranquility (-0.29) - all are of p<.001. The highest correlations between this dimension and a dimension of the facial expressions are with the happiness-sadness dimension (r=-0.28 p<.001) and with the surprise dimension (+0.23 p<.001). The content of this dimension is mainly of alertness (vigilance) and stubbornness versus serenity.

The third dimension - pride versus shame (embarrassment)

The marginal-words are pride + firmness + haughtiness versus fatigue + embarrassment. The highest correlations with words are: activity (+0.32); anxiety (-0.27); boldness (+0.41) courage (+0.43); decisiveness (+0.31); embarrassment (-0.31); fatigue (-0.34); firmness (+0.53); habitualness (-0.25); pride (+0.46); shame (-0.25); sleepy (-0.27); slumber (-0.25); superiority +0.42) and weakness (-0.27) - all are of p<.001.
Its highest correlations with subgroups are with sadness (-0.20, p<.02); with shame (-0.14, p<.05) and anger (+0.19, p<.02).

The significant correlations between it and dimensions of the facial expressions are with the anger dimension (+0.29 p<.001) and with the shame dimension (-0.19 p<.007).The content of this dimension is mainly of boldness, firmness and pride versus embarrassment, fatigue and shame.

The fourth dimension - surprise (and excitement) versus indifference

The marginal words are indifference + boredom + habitualness versus surprise + adoration curiosity + shame + panic.
The highest correlations with words are: boredom (+0.24); habitualness (+0.16); indifference (+0.42) and sleepy (+0.16) versus admiration (-0.39); excitement (-0.46); fondness (-0.40); gratitude (-0.48); love (-0.43); panic (-0.44); shame (-0.43) and surprise(-0.59).

-72-

Its highest correlations with subgroups are with surprise and fear of the free grading task (p<.002) and with surprise (-0.22 p<.002); contempt (-0.18 p<.02); sadness (-0.15 p<.05) and shame (-0.15, p<.05) in the Q-Sort task. The content of this dimension is mainly of excitement versus blandness.

The fifth dimension - the dimension of Pity

The marginal words are pity + compassion + leniency versus sympathy + curiosity + surprise + helpless.
The highest correlations (and mathematically negative - the same direction as that of pity) with words are with: belonging (-0.33); boldness (-0.38); cautious (-0.32); compassion (-0.53) conciliated (-0.32); firmness (-0.31); generosity (-0.34); hope (-0.33); intensity (-0.37); leniency (-0.56); pain (-0.30); pity (-67); tenderness (-0.41); tolerance (-0.30) - all are of p<.001. Only two words are positively correlated to it - confusion (+0.15 p<.038) and love (+0.17 p<.018). This result is also an indication of the imperfection of our sampling of the emotional domain.

The only significant correlation with a subgroup is with interest ( 0.18, p<.02). Besides its negative correlations with artificial facial expression items, it is also positively correlated with mixtures of anger + disgust and anger + contempt(both of r=+0.21, p<.004).

The fifth dimension is also significantly correlated with the second dimension of the facial expressions whose content is of positiveness and superiority (leniency) attitudes to the others (0.15 p<.036). The content of this dimension is mainly of pity, leniency and compassion.

The sixth dimension - longing versus skepticism (and astonishment)
The marginal words are longing + compassion + leniency versus astonishment + suspiciousness + skepticism + curiosity + caution + serenity. The placement here of serenity seems to be a result of the oddities of the S.S.A.-I program).

-73-

The highest correlations with words are with: compassion (-0.29); courage (-0.28); fatigue (-0.27); longing (-0.48); yearning(-0.27); worry (-0.25) versus astonishment (+0.35); skepticism (+0.24) and suspiciousness (+0.34). This dimension scores are significantly correlated with the fifth and sixth dimension of the facial expressions: the astonishment direction of this dimension as that of surprise of the fifth; the longing direction of this dimension is in the same direction as that of worry and guilt of the sixth. The content of this dimension is mainly of longing versus skepticism. [As the majority of the subjects were single and far from home this dimension and feeling may represent a basic motion of attachment (belonging) versus detachment (loneliness)].

The seventh dimension - responsibility and involvement

The marginal words are routine versus indifference + longing + skepticism. The highest correlations are with the words: animosity (-0.27); attraction (-0.30); delight (-0.30); indifference (-0.29); longing (-0.31); routine (+040); mirth (-0.24) and worry (+0.25).

It is significantly correlated with the subgroup of surprise (-0.15, p<.05). There are positive correlations with basic emotion items of distress no. 11 (+0.15 p<.05) and the low validity disgust no. 32 (+0.19, p<.02).

There is a significant correlation between it and the surprise dimension (of the facial expressions - (-0.14 p<.05) and a near significant one with the disgust dimension (+0.14 p<.055). The content of this dimension is mainly of the dreariness of the daily obligations versus fun.

The eighth dimension - patience and loneliness versus longing and love

The marginal words are hope + longing + excitement + desire + delighted versus loneliness + patience + involvement.

-74-

The highest (negative) correlations - with the same direction as that of longing (of p<.002) are with: happiness, craving, excitement, weakness, desire, alarm, slumber, fatigue, longing, and hope. The correlations with the other direction are of involvement, patience (0.31 p<.001) and leniency (0.33 p<.001).

The significant correlations with artificial facial expressions are: surprise no. 26 (-0.20 p<.004); one mixture of enjoyment + anger and one of disgust + fear (both of r=-0.19, p<.02) - of the free grading and with anger no. 28 (-0.18 p<.02); disgust + hesitation (-0.17 p<.02); disgust + fear (-0.23 p<.002); sadness + anger (+0.16p<.02); sadness + enjoyment (+0.18 p<.02) - of the Q-Sort procedure.

There is a negative correlation (of -0.19, p<.006) between this dimension and the anger dimension of the facial expressions.
The content of this dimension is mainly of patience (of the abstainers) versus the cluster of emotions of those preoccupied with love relationships.

The ninth dimension: guilt and depression versus contempt and disregard

The marginal words are caution + fatigue + disregard versus loneliness + contempt + skepticism.

The highest correlations with words (p< .002) are for: contempt, loneliness and skepticism versus caution, disregard, embitterment and fatigue. Following (with .002>p<.02) are boredom and involvement versus anger, confidence, blur, despair, guilt, haughtiness, imposer, pain, restlessness, servility, sleepy and slumber.

A possible common denominator for the words that are in the direction of caution and disregard is the depression syndrome - guilt, despair and pain, and their results - restlessness and fatigue.

The most significant correlation with artificial facial expressions is that of contempt no. 7 (+0.20 p<.004) of the free grades. There is a correlation of 0.18 (p<.012) between the guilt direction of the ninth dimension and the guilt direction of the sixth dimension of the facial expressions.

-75-

The content of the ninth dimension is mainly about the introjected contempt - guilt and its "relatives" versus its extrojection.

The tenth dimension - hostility versus conciliation

The marginal words are conciliated + serenity versus hate + pain + fatigue. The words that are of the p<.002 correlation with this dimension are all of the conciliated direction: alarm, alertness, adoration, conciliated, contentment, desire, disappointment, longing, repulsion, right, serenity.

The words whose correlation is in the direction of the hostility are: annoyance, boredom, fatigue, hate, pain, yearning - all of .002 On the relations between the ten dimensions of the words and those of the facial expressions The following Table 12 contains the matrix of correlations between the ten dimensions of the words and the ten dimensions of the 48 unmanipulated facial expressions. The mathematical directions of the results were adjusted in order to clarify the presentation.

***Table 12 about here***

The matrix of 100 correlations between the two sets of 10 dimensions contains 10 with p<.01 and among them 6 are of p<.001. There is a clear convergence of the first dimension of each set but not identity. It seems that in the universe of emotion there is a slightly different direction for each of them. This phenomenon is the root of the significant correlation of each of the two with the second dimension of the other set.

As the two sets of items that are the basis of each multidimensional analysis sample the emotional domain in a different way; and as the mathematical procedure of the analysis is dealing with variance that is, partly, not systematic - it is not realistic to expect a perfect fit between the two sets of due to the power of the new methodology, there is a significant convergence between six of the dimensions of the facial expressions, and eight of the dimensions of the 96 words of emotion.

-76-
Table no. 12: The matrix of correlations between the ten dimensions of the
words and the ten dimensions of the facial expressions
_________________________________________________________________________
Dimension   |  1  |  2  |  3  |  4  |  5  |  6  |  7  |  8  |  9  | 10  |
 *Facial    |Happi|Inter|  F  |  A  |  Su |Troub|  S  |  Co |Love |Disgu|
   *express |ness-|  est|  e  |  n  |  rp |led -|  h  |  nt |  _  |st-Sa|
     *  ions|Sad  |in ot|  a  |  g  |  ri |Light|  a  |  em |     |tisfa|
 Words   *  | ness| hers|  r  |  er |  se | ness|  me |  pt | Hate|ction|
------------------------------------------------------------------------|
1 Distress -| 0.60| 0.25|-0.12| 0.14|-0.04|-0.08|-0.10| 0.02| 0.13|-0.01|
 Contentment|0.001|0.001|0.094|0.044|  *  |  *  |  *  |  *  |0.065|  *  |
------------------------------------------------------------------------|
2 Alertness |-0.28| 0.10|-0.13| 0.08| 0.23| 0.17| 0.05| 0.00| 0.12| 0.05|
  - Serenity|0.001|  *  |0.060|  *  |0.001|0.018|  *  |  *  |0.097|  *  |
------------------------------------------------------------------------|
3 pride  -  | 0.01| 0.06| 0.01| 0.29| 0.05| 0.01|-0.19| 0.09|-0.03|-0.04|
Embarrassment| *  |  *  |  *  |0.001|  *  |  *  |0.007|  *  |  *  |  *  |
------------------------------------------------------------------------|
4 Excitement|-0.06|-0.07|-0.09| 0.00| 0.00|-0.07| 0.05|-0.10|-0.04| 0.07|
Indifference|  *  |  *  |  *  |  *  |  *  |  *  |  *  |  *  |  *  |  *  |
------------------------------------------------------------------------|
5 Pity      | 0.04| 0.15|-0.01|-0.03| 0.05|-0.06| 0.03| 0.02| 0.10|-0.10|
            |  *  |0.036|  *  |  *  |  *  |  *  |  *  |  *  |  *  |  *  |
------------------------------------------------------------------------|
6 Longing - | 0.12|-0.11| 0.06| 0.02|-0.16| 0.24| 0.08| 0.01| 0.04|-0.03|
  Scepticism|0.093|  *  |  *  |  *  |0.022|0.001|  *  |  *  |  *  |  *  |
------------------------------------------------------------------------|
7 (Ir)respon|-0.07| 0.02|-0.02| 0.05| 0.14|-0.08|-0.03|-0.05| 0.04| 0.14|
sibility    |  *  |  *  |  *  |  *  |0.050|  *  |  *  |  *  |  *  |0.055|
------------------------------------------------------------------------|
8 Patience  |-0.05| 0.09| 0.01|-0.19| 0.00|-0.01| 0.05|-0.03|-0.04| 0.04|
     -  Love|  *  |  *  |  *  |0.006|  *  |  *  |  *  |  *  |  *  |  *  |
------------------------------------------------------------------------|
9 Guilt  -  | 0.05|-0.05|-0.02|-0.05| 0.04| 0.18|-0.06| 0.04|-0.01|-0.08|
    Contempt|  *  |  *  |  *  |  *  |  *  |0.012|  *  |  *  |  *  |  *  |
------------------------------------------------------------------------|
conciliation| 0.09| 0.11| 0.08| 0.08| 0.03|-0.09|-0.03|-0.02| 0.05|-0.10|
10 Hostility|  *  |  *  |  *  |  *  |  *  |  *  |  *  |  *  |  *  |  *  |
=========================================================================
* Significance is of P>0.10
-77-

There is a significant converging between the common content of the two sets of dimensions, and the basic emotions of the evolutionary theories. Even most of the content that seems to be pronounced clearly only in one set of dimensions, converges with basic emotions of other studies.

Summary of the results of the examination of the second hypothesis

The results do not support the second hypothesis of this study and thus refute our theoretical effort to reconcile the "parsimonious"-cognitive approach and the evolutional approach.

Our results show clearly that the dimensions of discrimination among words of emotion have a concrete content that fit results of previous studies about basic inborn emotional variables. Our study undermines the firm belief in the "parsimonious" models of two or three dimensions that were offered for the emotional domain. It seems that methodological explanations - which are presented in the following chapter - are in more accord with our results. This explanation conforms to the criticism of Ekman, et al. (1982) about the methodology of research in the field of facial expression of emotion. It is also more parsimonious than the theory we offered as it does not call for two sets of emotional variables.

To get back ON-LINE with the Biofeedback WITHOUT instruments
The other Home Pages: http://www.netvision.net.il/php/gshalif
http://www.geocities.com/~drilanshalif
The E-Mail: gshalif@netvision.net.il
To the TOP

For suggestions and complains click on: gshalif@netvision.net.il

(=====Back to result I.
To the Discussion ---->