Social acts or joint actions are an important aspect of human life. Joint
actions are activities done by at least two people or more. For instance
dialogues, teamworks, two(or more)-player games, dancing, and so on.
Coordination and harmony, are essential componens of joint actions (Sebnaz
& Knoblich, 2009). People need to act proportionate to their partner’s
actions for a normal joint action.
There is several theory trying to explain joint actions mechanisms and
components. Here we follow Pickering and Garrod (2013) grouping of theories called
“classic theories”. Classic theories of joint actions contain dominant idea
which we called sensory approach to understanding of other’s actions or
intentions. These theories assumed perceiver collect sensory inputs, then
process and decode massages. Classic theories suppose serial approach to
production (action) and perception (action perception) of massages in joint
actions. In other words, classic theories propose aggregative and bottom-up
perception for understanding of others.
Some featurs of joint actions and some new findings, questioned classic
theories. Joint actions contain speed and fluency normally, which make serial
approach unapproprate to explain it; it seems only sensory aggregation and
decoding be a slow mechanism for what happen normally in joint actions.
New findings such as mirror neurons, propose action and action perception
are interwoven structurally and functionally (reference), The interwoven
structure of action and percepion propose that understanding of other’s
behavior done through active simulation rather than sensory agreggation. Understanding
of other’s movements (wolpert, 1998, 2003) and their utterance in dialogue
(Pickering & Garrod, 2013).
Mirror neurons: mirror neurons studies suggest perception of individual’s
actions is active reproduction of the actions(Maldonat & Delorco, 2013). In
other words, action and action perception, subordinated to same neurons (Mukamel,2010;
Maldonato, 2013; Oshi, 2017; Waitz, 2011). We focus on findings propose mirror
neurons activation is not only during the observation of actions but it is before
actions, suggested they work predictive.
Single neuron recording on Macaque monkeys showed when individual’s action
hide from animals in a context, mirror neuron activate too. So it seems mirror
neurons playing role in predivtive inferring (Omilta ef al, 2001). Single pulse
TMS and motor evoked potential on professional and amateur basketballists
showed prediction of an individual’s hands movements, activate the same muscles
in observers (Ferrari and Rizolati, 2015, p 99?). Another study illustrated the
role of prediction in skiing (Bertoz, 1997). The findings suggest a new
component in the mechanism of understanding of other’s behaviors in joint
Wolpert: Wolpert (1998, 2003) suggested computational model proposed for motor
control (MOSAIC) can be developed to social interaction sphere. Wolpert mentioned
some points why aggregation of sensory feedbacks, can not be an appropriare
mechanism to motor control. Major point is sensory signals arrive to CNS by
delay (some milliseconds). Moreover, sensory inputs suffer from neural noise,
so motor control must be limitted to perform rapid and accurate movements only
by sensory information (Harris & wolpert, 1998). The delay and noise, increase
more in social interactions. Delay increase to few seconds or much more (Wolpert,2003).
So sensory feedback aggregation couldn’t explain velocity and fluency of social
acts (Pickering & Garrod,2004). Thereupon, both motor control and social
interactions need to work through some sort of prediction before getting
sensory feedback. Wolpert (2003) expressed multiple contoller exist
simultanously to motor control. To choose optimum ones, individuals intend to
contexts and states. Every cotroller contains a “forward model” or an internal
model represented relationship between actions and their sensory effects. So it
makes predictions possible to individuals. Thus people can modify their
behavior before end of their actions and getting sensory feedback. Wolpert
developed the Idea from motor control sphere to social interactions and other’s
movements understanding, reffering to mirror neurons findings. Mirror neurons
findings propose that people understand other’s actions through activating their
own action system at a sub-threshold level. As we told before, findings show
this activation works based on prediction. people have predicting Idea about
other’s next behavior befor observing the real actions. So Wolpert could develop
prediction and Forward model concept to understanding of other’s actions in
Pickering & Garrod: Pickering and Garrod (2013), developed wolpert’s model to dialogue as some sort of social
interactions. They gathered many evidence at the different levels of language (semantics,
syntax and phonetics) that suggest people don’t wait for sensory input of other’s
utterance but they make prediction of what they will perceive (references).
Wolpert highlighted the importance of prediction in his proposed model; mirror
neurons was the experimental source of inspiration of the model. For this
reason, the only mechanism of prediction was considered “simulation” by Wolpert.
Simulation means reproduction of the observing action by perceiver. Pickering
and Garrod declared simulation is not the only way which makes predictions
possible. People could make prediction through prior experiences and
observation with some one. They called this way “assosiation route” of
prediction against simulation route. Anyway, Pickering and Garrod just payed
attention to simulation route.
Here we intend to focus on assosiative route of prediction: predicting
based on prior experiences and habits.
ERN: event-related potential studies have introduced an electrophysiologic
peak correlared to errors, called Error-related negativity (ERN) most of the
time. ERN time span was reported variable according to experimental design but
it usually occures between 30 to 80 ms after making an error. ERN doesn’t correlate
to awareness of errors necessarily. Some functional theory of ERN correlated it
to “error detection”. Another theory propose ERN functioning as a “comperator”
in conflicting and inconsistent conditions. Conditions can be conflicting when
individuals observe difference between what they have predicted and what is
currently perceiving. The recent theory is more general and can encompasses
first theory (error detecting as comperating desired outcome to real outcome
and finding difference between them). ERN is the outcome of reducing Dopamin
output to cortex, paticularly Anterior Cingulate Cortex (ACC) neurophysiologicly.
When individual’s behavior is worth than his/her anticipation, entrance of
Dopamin reduce to cortex. So ACC neurons would be inhibited and depolarized. Reduction
of Dopamin amount in cortex, reduce inhibition of ACC neurons and leads to a
negative signal called ERN.
New findings suggested ERN is representation of not only committing error but
it is related to more general events like as failure to predicting things
(references in p 43 thesis). Some researches asserted the role of individual’s prediction
and anticipation on ERN; Less anticipation for being wrong, more emplitude of
ERN (Holroyd et al, 2003). It seems being non-intentional and unpredictable is
a main factor in ERN. When people do error intentionally, ERN doesn’t occure
(stemmer, Witsky, Sconnel, 2001). Moreover, some studies show ERN is not a
signal correlated to making an error really, but it can be detected before
compeleting the action (Richardson,… 2013). Above all, ERN is not just
correlated to sombody’s own error, but to only observation of other’s error or
just what is conflicting to his/her prediction (for example Newman 2009, Ma,
2011). In other words, people make prediction both about their own actions and
consequences and other’s behaviors at defferent levels. Chi and et al (2014)
showed ERN can be detectable before observing others compeleting their error
actions. ERN was detected just by observing the onset of action is going to be erroneous
(references p 48,49 thesis). It seems ERN is detectable for mismatch between a
predicted consequence, desire or wish and what really occurred. People are
predictor; They predict what are communicating to; they are not only aggregator
for sensory inputs and feedbacks in either their action or other’s actions.
Utilization of pediction to understanding of other’s behaviors can explain
speed and fluency of joint and social actions.
In present research, according to Wolpert (1998,2003) and Pickering and
Garrod (2013) models, we studied does individual’s understanding of other’s behavior is based on prediction?
Moreover, as wolpert expressed, sensory synaptic weights will be modified after
each prediction error and it leads to improve forward model’s predictions. In other
words, prediction errors lead to an updating on internal models making system’s
prediction optimum in the future. We propose study the updating experimentally
and explore behavioral and electrophysiologic correlates of it.