Measures of Emotions in SocLab

Following the OCC structure of emotions suggested by Ortony et al. (1988), TABLE I shows the emotions a SocLab actor is likely to feel in a configuration of the organization. The occurrence and intensity of emotions are given in pairs, defined as a potential (given in terms of a proportion) and two thresholds. If the potential is above the high threshold then the positive emotion is likely to occur, and if it is below the low threshold then the negative emotion is likely to occur. In case the potential is between the thresholds then no emotion raises. The proportion is always a comparison between what actually happens (e.g., the influence exerted by the actor) and what could happen (e.g., the range of influence it could exert). Indeed, a social actor “appraises” the situation in the context of the possibilities available for it. The emotional interpretation of the values of each index depends on the very nature of the organization under consideration and of individual traits of the actor a. Globally, considering as an example the Joy/Distress emotion, one could consider that Joy appears above 70% (high threshold) and distress under 50% (low threshold).

These indexes are not variables used by SocLab actors in their decision making process. They are based on essential properties of configurations, i.e., what is given (Influence) by a to b, and what is received (Satisfaction) by a from b, where a and b may be: a particular actor, or the whole organization (SOA), as shown in Table I. Below, we will use short names for the variables: Sat(a,s) for satisfaction(a,s), Inf(a,s) for influence(a,s), minSat(a) (resp. maxSat(a)) for the minimal (resp. maximal) satisfaction a can receive from the whole. For a more detailed presentation, see Terán and Sibertin-Blanc (2014).

Table I.  Emotions Experienced by an actor in SocLab

measuresOfEmotionsTable1

i.  Well-being emotions: Joy/Distress

The OCC model defines joy (resp. distress) as to be pleased (resp. displeased) about the occurrence of a desirable (resp. undesirable) event. In the SocLab model of an organization, the occurrence of such an event corresponds to reaching a regulated configuration that is satisfying (resp. dissatisfying). Joy/Distress of an actor a is given as:

Potential:

propSat(a,s)=(Sat(a,s) – minSat(a))/(maxSat(a) – minSat(a)),   which value is in [0, 1].

Thresholds: JoyThresh(a) and DistressThresh(a) are the maximal (resp. minimal) values of the proportion of satisfaction a receives generating the emotion Joy (resp. Distress). These thresholds make the actor liable to these emotions.

Intensity:    Joy(a, s) = max{0, propSat(a, s) – JoyThresh(a)}
Distress(a,s) = max{0, DistressThresh(a) – propSat(a,s)}.

Other case of Joy/Distress emotion is related with the SOA’s global satisfaction, and a’s group identification with the SOA, which has intensity:

Joy(a, s) =  max{0, (propGSat(s) – JoyGThresh(a)) * GI(a)}

Distress(a,s) =max{0,(DistressGThresh(a)-PropGSat(a))*GI(a)}.

ii. Attribution emotions: Pride/Shame

An actor could feel prideful (resp. guilty or shameful) when he approves (resp. disapproves) its own praiseworthy (resp. blameworthy) action regarding its effect on itself, or on some other actor(s) close to it. But an actor can also feel prideful by a praiseworthy action performed by another actor close to it. This proximity of an actor a toward another actor b can evaluated as its Cognitive Unity cogUnit(a,b) ∈ [-1, 1]. Thus, the pride/shame of a when it evaluates what b gives to c will be (it might be the case that either a=b, a=c or b=c):

Pride(a, b, c, s) = max{0, (propInf(b,c,s) – PrideThresh(b, c)) * cogUnit(a, b) * cogUnit(a, c)},

Shame(a,b,c,s) = max{0,(ShameThresh(b, c)-propInf(b, c, s)) * cogUnit(a, b) * cogUnit(a, c)},

provided that cogUnit(a,b) > 0and cogUnit(a,c) > 0, otherwise both pride and shame are null, and where:
propInf(b,c,s) = (Inf(b,c,s)-minInf(b,c)) / (maxInf(b,c)-minInf(b,c))

So, we define pride/shame as the product of a measure of actor a’s approval of the action of b (profInf(a,b,s)), multiplied by the cognitive units of a with b and with c. This is in accordance with the extension of OCC proposed by Steunebrink (2012) (they consider only the case a = c).

iii. Attribution emotions: Admiration/Reproach

Admiration (resp. reproach) is related to actor a approving (resp. disapproving) another actor b’s praiseworthy (resp. blameworthy) action, towards a third actor c (that can be the SOA or a particular actor). In this case, a sees b as another, i.e. cogUnit(a,b) is null, it has a positive cognitive Unit towards c and it evaluates the influence of actor b on c. Thus,the admiration/Reproach of a towards b when cogUnit(a, b) = 0 is defined as

Admiration(a,b,c,s) = max{0,(propInf(b,c,s)-AdmThresh(b,c))*cogUnit(a, c)},

Reproach(a,b,c,s) = max{0,(ReproachThresh(b, c)-propInf(b,c,s))*cogUnit(a, c)}.

iv. Well-being/Attribution compounds emotions: Gratification/Remorse

OCC defines gratification (resp. remorse) as being pleased (resp. displeased) about a desirable (undesirable) event or situation that results from oneself action and thus entails the approving (resp. disapproving) of one’s own praiseworthy (resp. blameworthy) action. As said above, an event is related with the action of the whole SOA, which results from action of individuals. If the actor feels Joy (resp. Distress) about the situation of the SOA and it considers himself as responsible for it, then it will feel Gratification/Remorse, as follows:

Gratification(a, s) = max{0,(propGSat(s) – GratifThresh(a))* GI(a) *propInf(a, s)},

Remorse(a, s) = max{0,  (RemorseThresh(a) – propGSat(s)) * GI(a)* proInf(a, s)}.

v. Well-being/Attribution compounds emotions: Gratitude/Anger

The OCC model defines gratitude (anger) as to be pleased (displeased) about the consequences for oneself of another’s praiseworthy (blameworthy) action. Thus gratitude (anger) is similar to gratification (remorse), but it regards what is given by someone else instead of what is given by oneself. We define the Gratitude/Anger of a towards b as:

Gratitude(a,b,s) = max{0,(propGSat(s)-GratitThresh(a))*GI(a)* propInf(b, s)}

Anger(a, b, s) = max{0, (AngerThresh(a) – propGSat(s)) * GI(a)* propInf(b, s)}.

vi. Fortunes-of-others emotions: Happy-for/Resentment, and Gloating/Pity

These emotions appear when the actor perceives what is happening for another particular actor as a consequence of a configuration resulting from collective action. As an example, when an actor b is distressed because he is receiving less than the minimal he wishes. Under this situation, if actor a has negative (resp. positive) solidarity towards actor b, a feels pleased (resp. displeased) by what is happening to b, and so would experience the potential for gloating/pity (resp. Happy-for/resentment) as follows:

  • in case undesirable event for b, i.e., Distress(b, s) < 0:

gloating/pity(a,b,s) = abs(Sol(a,b) * Distress(b,s))  (pity if Sol(a,b) > 0, gloating if Sol(a,b) < 0)

  • in case desirable event for b, i.e., Joy(b, s) > 0:

happy-for/resentment(a,b,s) = abs(Sol(a,b) * Joy(b,s))  (happy-for if Sol(a,b) > 0, resentment if Sol(a,b) < 0)

A case study: The TDPM Team

A model of a concrete team is used to exemplify how social emotions can be identified in SocLab, and to illustrate how such identification can help in auditing organisations or designing policies for promoting collaboration

Example

We show in TABLE II and TABLE III the potential of Joy of all actors, for Group Identification (GI) of some actors in [0, 1] (GI is varied with step size of 0.1). The GI of the other actors is kept in 0. In TABLE II the actors whose GI is varying are: the Director, researcher_W, developer_W, and techSpreader_W, i.e., all those actors weakly engaged/identified. In order to check the effect of varying only the GI of the Director, only this actor’s GI is varied in TABLE III.

We can see from TABLE II that the potential of Joy of the Director decreases from 100 to 68, while the emotions of the other non strongly engaged actors (lines in blue) increase, and the emotions of the engaged actors do not suffer significative changes. Also we see how collaboration of the Director suffers a significative change as the state of the relation it controls decreases from 2.5 to -8.2. The state of the relations controled by the other weakly engaged actors do not change significatively.

TABLE III shows that the main effect on the joy of actors come from the variation of GI of the Director. This result is useful to orient management decisions to increse collaboration and joy of actors in the TDPM team. Additionally, TABLE III shows that collaboration of the actor techSpreader_W decreases, as the value of the relation it controls decreases, going from -8.5 to -9.9, taking advange of the Director higher collaboration when its GI increases.

TABLE II. Potential of JOY felt by the actors of the TDPM team, and state of the relations, FOR GI=0, 0.1, .., 1 for actors: Director,researcher_W, developer_W and techSpreader_W (lines in blue). The GI of the other actors remain in 0.

measuresOfEmotionsTableII

TABLE III. Potential of JOY felt by the actors of the TDPM team, and state of the relations, FOR GI=0, 0.5,0.8, and 1 for the Director (line in blue).The GI of the other actors remain in 0.

measuresOfEmotionsTableIII

References

Ortony A., G. Clore, and A. Collins. 1988. The Cognitive Structure of Emotions. New York: Cambridge University Press.

Steunebrink, B., M. Dastani, and J.-J.Ch. Meyer. 2012. A formal model of emotion triggers: an approach for BDI agents. Synthese. 185(1):83-129. Springer.http://www.idsia.ch/~steunebrink/Publications/SYNTHESE_emoti

Terán Oswaldo and Christophe Sibertin-Blanc. 2014. Indexes of Emotions in Organisational Settings.  submitted.