Do We Click? Speaker-listener
        neural coupling underlies
       successful communication
 Lauren Silbert1, Greg Stephens2,3, Uri Hasson1

   1PrincetonNeuroscience Institute and Psychology Department,
Princeton University, 2Lewis-Sigler Institute for Integrative Genomics,
  and 3 Joseph Henry Laboratories of Physics, Princeton University,
                            Princeton, NJ
Verbal communication is a joint activity through
      which individuals share information




              Listener        Speaker
          (comprehension)   (production)
Communication is integral to the functioning of
          our complex societies




 Globalization and the rapidly changing world
demand we increase our understanding of how
          communication succeeds
Verbal communication is a joint activity through
      which individuals share information




              Listener        Speaker
          (comprehension)   (production)
Neurolinguistic studies constrained by
   boundaries of individual brains




         Listener        Speaker
     (comprehension)   (production)
Neurolinguistic studies constrained by
   boundaries of individual brains




         Listener        Speaker
     (comprehension)   (production)
Assessing an interactive process:
laboratory verse real-world stimuli

          Control simplified stimuli

                 red
                 cat
                 bat




 Isolation from the environment
Assessing an interactive process:
laboratory verse real-world stimuli

          Control simplified stimuli

                 red                   Complex natural stimuli
                 cat
                 bat




                                        Two brains interacting

 Isolation from the environment
Design

     Speaker



  15 min monologue




               (real life unrehearsed story)
FOMRI™ II Dual Channel MRI Microphone System
                           Mic 1

                          Mic 2


                    Two orthogonal pressure gradient optical microphones




Mic 1: Signal +
         noise


   Mic 2: noise


     Extracted
        signal
Design

     Speaker                                   Listener



  15 min monologue
                                                 (n=11)




               (real life unrehearsed story)
Brain activity during production as a model for
                comprehension
 model




                                                  (n=11)
Brain activity during production as a model for
                comprehension
 model




                                                  (n=11)
Brain activity during production as a model for
                comprehension
 model                       Listener’s response




                                                   (n=11)
Brain activity during production as a model for
                comprehension


•  Bypass the need to specify a priori any formal model
of linguistic processes in any given brain area

•  Ability to look at similarity of responses during speech
production and speech comprehension in the same area
Adding Temporal
Dynamics to the Model:




                         (n=11)
Adding Temporal
Dynamics to the Model:




                         (n=11)
Adding Temporal
Dynamics to the Model:




                         (n=11)
Adding Temporal
Dynamics to the Model:




                         (n=11)
Adding Temporal
Dynamics to the Model:




                         (n=11)
Adding Temporal
Dynamics to the Model:




                         (n=11)
Adding Temporal
Dynamics to the Model:




                         (n=11)
Adding Temporal
Dynamics to the Model:




                         (n=11)
Adding Temporal
Dynamics to the Model:




                         (n=11)
Conservative Statistics: Parametric and Non-parametric
                                  Correct for
  Analytic F-test applied         multiple
  to overall model fit            comparisons
                                  using FDR
                                  g=0.05      1000




                            Actual correlation
                               coefficient




                                                     (n=11)
Predictions




1.  Are neural responses during speech production
    and speech comprehension coupled?

2.  Can the extent of speaker-listener neural
    coupling predict the success of communication?
Predictions




1.  Are neural responses during speech production
    and speech comprehension coupled?

2.  Can the extent of speaker-listener neural
    coupling predict the success of communication?
Brain responses are reliably shared across all
                  listeners during comprehension
response




                                                        subject 4
  f MRI




                                                        subject 3
                                                        subject 2
                                                        subject 1

                                time
Speaker’s brain responses during production are
coupled to listener’s responses during comprehension
Areas involved in comprehension overlap with areas
          coupled during communication
Necessary controls


I.    Is the speaker-listener coupling tied to the
       content of the story?
         •  Block Communication
         •  Misaligned Communication
Speaker-Listener neural coupling is absent in the
    absence of successful communication




                                         ?

                      X

  Russian Speaker               Non-Russian speaking
                                     Listeners
Speaker-Listener neural coupling is absent when
        communication is incongruous




                                        ?

                          X

Speaker telling story 2         Listeners to story 1
Necessary controls


I.    Is the speaker-listener coupling tied to the
       content of the story?
         •  Block Communication
         •  Shuffle Communication

II. Is the speaker a listener of herself?
       •  Temporal dynamics
Brain responses among listeners are time-locked to
            the moment of vocalization

   Average beta weights
Listener’s brain responses mirror the speaker’s
          brain responses with a delay

 Average beta weights
Brain responses shared among listeners are
temporally aligned to the moment of vocalization
Temporal dynamics of speaker-listener neural
       coupling varies across areas
Importance of temporal coupling results:

•  Dynamics of coupling between a speaker and a
listener are fundamentally different from dynamics
shared among all listeners.


                        vs



• Spatial specificity of temporal coupling
demonstrates effect cannot be attributed to non-
specific, spatially global effects like arousal.
Predictions




1.  Are neural responses during speech production
    and speech comprehension coupled?

2.  Can the extent of speaker-listener neural
    coupling predict the success of communication?
Level of understanding varies across listeners, as
    measured by reliable independent raters
The degree of neural coupling predicts the success
                of communication




        Speaker-listener Neural Coupling
Areas where the listener’s responses precede the
speaker’s show strongest correlation with behavior
Summary


I.    In the course of communication the listener’s brain responses
       become coupled with the speaker’s brain responses.
Summary


I.     In the course of communication the listener’s brain responses
        become coupled with the speaker’s brain responses.

II.    The extent of speaker-listener neural coupling is indicative of the
       success of the communication.
Summary


I.     In the course of communication the listener’s brain responses
        become coupled with the speaker’s brain responses.

II.    The extent of speaker-listener neural coupling is indicative of the
       success of the communication.

III.  An ability to evoke similar brain patterns in another individual via
      speech may gate our communication abilities.
Summary


I.     In the course of communication the listener’s brain responses
        become coupled with the speaker’s brain responses.

II.    The extent of speaker-listener neural coupling is indicative of the
       success of the communication.

III.  An ability to evoke similar brain patterns in another individual via
      speech may gate our communication abilities.

IV.  The recording of neural responses from both the speaker brain
     and the listener brain may be used to assess verbal and non-
     verbal forms of interaction in both human and other model
     systems.
Thank You
Christopher Honey   David Heeger
Yulia Lerner        Bruno Galantucci
Chris Thompson      Simon Garrod
Mina Cikara
Alana D'Alfonso
mirror neurons

Do We Click? - Laurent Silbert - H+ Summit @ Harvard

  • 1.
    Do We Click?Speaker-listener neural coupling underlies successful communication Lauren Silbert1, Greg Stephens2,3, Uri Hasson1 1PrincetonNeuroscience Institute and Psychology Department, Princeton University, 2Lewis-Sigler Institute for Integrative Genomics, and 3 Joseph Henry Laboratories of Physics, Princeton University, Princeton, NJ
  • 2.
    Verbal communication isa joint activity through which individuals share information Listener Speaker (comprehension) (production)
  • 3.
    Communication is integralto the functioning of our complex societies Globalization and the rapidly changing world demand we increase our understanding of how communication succeeds
  • 4.
    Verbal communication isa joint activity through which individuals share information Listener Speaker (comprehension) (production)
  • 5.
    Neurolinguistic studies constrainedby boundaries of individual brains Listener Speaker (comprehension) (production)
  • 6.
    Neurolinguistic studies constrainedby boundaries of individual brains Listener Speaker (comprehension) (production)
  • 7.
    Assessing an interactiveprocess: laboratory verse real-world stimuli Control simplified stimuli red cat bat Isolation from the environment
  • 8.
    Assessing an interactiveprocess: laboratory verse real-world stimuli Control simplified stimuli red Complex natural stimuli cat bat Two brains interacting Isolation from the environment
  • 9.
    Design Speaker 15 min monologue (real life unrehearsed story)
  • 10.
    FOMRI™ II DualChannel MRI Microphone System Mic 1 Mic 2 Two orthogonal pressure gradient optical microphones Mic 1: Signal + noise Mic 2: noise Extracted signal
  • 11.
    Design Speaker Listener 15 min monologue (n=11) (real life unrehearsed story)
  • 12.
    Brain activity duringproduction as a model for comprehension model (n=11)
  • 13.
    Brain activity duringproduction as a model for comprehension model (n=11)
  • 14.
    Brain activity duringproduction as a model for comprehension model Listener’s response (n=11)
  • 15.
    Brain activity duringproduction as a model for comprehension •  Bypass the need to specify a priori any formal model of linguistic processes in any given brain area •  Ability to look at similarity of responses during speech production and speech comprehension in the same area
  • 16.
    Adding Temporal Dynamics tothe Model: (n=11)
  • 17.
    Adding Temporal Dynamics tothe Model: (n=11)
  • 18.
    Adding Temporal Dynamics tothe Model: (n=11)
  • 19.
    Adding Temporal Dynamics tothe Model: (n=11)
  • 20.
    Adding Temporal Dynamics tothe Model: (n=11)
  • 21.
    Adding Temporal Dynamics tothe Model: (n=11)
  • 22.
    Adding Temporal Dynamics tothe Model: (n=11)
  • 23.
    Adding Temporal Dynamics tothe Model: (n=11)
  • 24.
    Adding Temporal Dynamics tothe Model: (n=11)
  • 25.
    Conservative Statistics: Parametricand Non-parametric Correct for Analytic F-test applied multiple to overall model fit comparisons using FDR g=0.05 1000 Actual correlation coefficient (n=11)
  • 26.
    Predictions 1.  Are neuralresponses during speech production and speech comprehension coupled? 2.  Can the extent of speaker-listener neural coupling predict the success of communication?
  • 27.
    Predictions 1.  Are neuralresponses during speech production and speech comprehension coupled? 2.  Can the extent of speaker-listener neural coupling predict the success of communication?
  • 28.
    Brain responses arereliably shared across all listeners during comprehension response subject 4 f MRI subject 3 subject 2 subject 1 time
  • 29.
    Speaker’s brain responsesduring production are coupled to listener’s responses during comprehension
  • 30.
    Areas involved incomprehension overlap with areas coupled during communication
  • 31.
    Necessary controls I.  Is the speaker-listener coupling tied to the content of the story? •  Block Communication •  Misaligned Communication
  • 32.
    Speaker-Listener neural couplingis absent in the absence of successful communication ? X Russian Speaker Non-Russian speaking Listeners
  • 33.
    Speaker-Listener neural couplingis absent when communication is incongruous ? X Speaker telling story 2 Listeners to story 1
  • 34.
    Necessary controls I.  Is the speaker-listener coupling tied to the content of the story? •  Block Communication •  Shuffle Communication II. Is the speaker a listener of herself? •  Temporal dynamics
  • 35.
    Brain responses amonglisteners are time-locked to the moment of vocalization Average beta weights
  • 36.
    Listener’s brain responsesmirror the speaker’s brain responses with a delay Average beta weights
  • 37.
    Brain responses sharedamong listeners are temporally aligned to the moment of vocalization
  • 38.
    Temporal dynamics ofspeaker-listener neural coupling varies across areas
  • 39.
    Importance of temporalcoupling results: •  Dynamics of coupling between a speaker and a listener are fundamentally different from dynamics shared among all listeners. vs • Spatial specificity of temporal coupling demonstrates effect cannot be attributed to non- specific, spatially global effects like arousal.
  • 40.
    Predictions 1.  Are neuralresponses during speech production and speech comprehension coupled? 2.  Can the extent of speaker-listener neural coupling predict the success of communication?
  • 41.
    Level of understandingvaries across listeners, as measured by reliable independent raters
  • 42.
    The degree ofneural coupling predicts the success of communication Speaker-listener Neural Coupling
  • 43.
    Areas where thelistener’s responses precede the speaker’s show strongest correlation with behavior
  • 44.
    Summary I.  In the course of communication the listener’s brain responses become coupled with the speaker’s brain responses.
  • 45.
    Summary I.  In the course of communication the listener’s brain responses become coupled with the speaker’s brain responses. II.  The extent of speaker-listener neural coupling is indicative of the success of the communication.
  • 46.
    Summary I.  In the course of communication the listener’s brain responses become coupled with the speaker’s brain responses. II.  The extent of speaker-listener neural coupling is indicative of the success of the communication. III.  An ability to evoke similar brain patterns in another individual via speech may gate our communication abilities.
  • 47.
    Summary I.  In the course of communication the listener’s brain responses become coupled with the speaker’s brain responses. II.  The extent of speaker-listener neural coupling is indicative of the success of the communication. III.  An ability to evoke similar brain patterns in another individual via speech may gate our communication abilities. IV.  The recording of neural responses from both the speaker brain and the listener brain may be used to assess verbal and non- verbal forms of interaction in both human and other model systems.
  • 48.
    Thank You Christopher Honey David Heeger Yulia Lerner Bruno Galantucci Chris Thompson Simon Garrod Mina Cikara Alana D'Alfonso
  • 50.