1. Presented By Meet Shah

2. Goal  Automatically predicting the respondent’s reactions (accept or reject) to offers during face to face negotiation by analyzing visual and acoustic behavior.  Just prediction and not to get high accuracy

3. Related work  General role of affect  Influence of mood  Personality  Emotions  Social context  Intuition based on affective and social perspective of negotiation and Studied employment negotiation and interview scenarios  Non verbal behavior can give clues

4. Non-verbal factor  Non verbal behavior of proposer /respondent [proxemics, body posture, gestures, facial expressions, para-language, emotional attitudes, hand movements, head shake, head tilt]  Mutual behavior symmetry / asymmetry [behavior matching, imitation [speech rate, poses, accents, tone of voice], mimicry, synchrony, chameleon effect, eye contact, forward lean, gesture / posture shifts ]  History [past history between negotiators]

5. Computational Desciptors  Extra feature of time dependency  Short term : [cheating]  long term cues : [Mutual gaze]

6. Proposer's behavior  The proposer’s behavioral cues [Head node, head shake, head tilt, gaze, smile, self touch] were manually annotated within the time window of each proposal-response event and were encoded as binary descriptors at the event level.  For example : smile of proposal from start to end

7. Respondent's behavior  Binary response time: For each proposal-response event, the response time was computed as the time when the respondent started uttering acceptance or rejection minus the time when the proposer finished uttering his/her proposal. After taking the means of the response times for all accepted and for all rejected cases, the midpoint of the two means was found and used as a threshold, which was 1.37 seconds in our experiments. Using this threshold, the response time in each proposal-response event was converted into a binary descriptor.

8. Mutual behavior  Behavioral symmetry [mutual gaze or reciprocal smile]  Behavioral asymmetry [Contrast between behavior their behavior]

9. Negotiation history  Net negotiation history : The total net response history of the respondent at the time of the proposal response event (1 and –1 for each previous acceptance and rejection respectively).  Last negotiation history : The result of the proposal response event (þ1 for acceptance and –1 for rejection) immediately prior to the current one.  Response time history : The mean of all the previous response times of the respondent at the time of the proposal-response event. This descriptor could help better understand the binary response time descriptor by providing the general response time characteristic habit of each negotiator.

10. Experiment  Hypothesis 1 (H1). For predicting respondent reactions during negotiation, other sources of information (proposer’s nonverbal behavior, mutual behavior, and negotiation history) can yield comparable prediction performance to looking at nonverbal behavior of the respondent, and combining all sources together yields higher performance than using a single source of information.  Hypothesis 2 (H2). Computational descriptors of mutual behavior that are predictive of respondent reactions are also useful for determining whether the negotiation interaction is cooperative or competitive.

11. Dataset  42 face to face negotiation sessions each involved same sex participants to control the influence over gender  84 undergraduate students [40 males, 44 females]  A total of three cameras were placed unobtrusively to record a near-frontal view of each negotiator, as well as an overall side view of the interaction  Randomly assigned restaurant where they have to negotiate how to organize and distribute fruits and vegetables on the table.

13. Prediction Model and Methodology  four-fold cross-validation was performed with hold-out testing and also hold-out validation to find the optimal parameters (gamma and C) using a grid-search technique.  In order to make balanced sample sets for predictor training and testing, all of the 63 samples of the rejected proposal - response events were combined with 63 randomly selected samples of the accepted events to make the 50% baseline [forced balance and universal for all measures].  For the 2 nd hypothesis, The samples were also randomly balanced with 13 cooperative sessions and 13 competitive sessions (making the baseline classification at 50 percent), and similar feature selection technique and 13-fold cross validation [8 –training and 5 - validation]was performed using same descriptors.

16. Conclusion

17. Questions  1. Would the results of the second hypothesis differ if one participant was told to be cooperative and the other to be competitive?  2. The study from this paper is hard to materialize to real world situations including "application of training a person to be a better negotiator" as they do not study intricacy of opposite gender dyad, where the flow of influences are more complex than same gender pairing.  3. “What other social ques the authors could have taken into their consideration apart from what they have used in their paper for both the proposer and respondent.”  4. The authors find that acceptance and refusals occur together in blocks, and they also take negotiation history into account while predicting the response. Does this mean that a “yes” after a group of “no” would be wrongly predicted?

18. Questions  5. Do the results depend on the “quality” of the dataset used, or it can be generalized on real world scenarios without significant reduction of accuracy?  6. What kind of technique and how authors have used to scale all mutual behavior? [Ex. They have given range from intensity 0 to 100 to smile of the interactants.]  7. How did the authors capture predictive cues regarding negotiation history of proposer and respondent?  8. What about the influence/impact of factors such as mood,personality etc on a persons non verbal behaviour ?