Linear readout and its interpretation.

<p>(A) We study a “standard” model of percept formation, with two parameters <i>w</i> and <i>t</i>_<i>R</i> defining integration in time, and a readout vector <b>a</b> defining integration across neurons. (B) Geometric interpretation of the model. The temporal parameters <i>w</i> and <i>t</i>_<i>R</i> define the tuning vector </p><p></p><p></p><p>b<mo>‾</mo></p><p></p><p></p> and noise covariance matrix <p></p><p></p><p></p><p>C<mo>‾</mo></p><mo>‾</mo><p></p><p></p><p></p> in the population. Colored ellipses represent the distribution of neural activities from trial to trial, for the three possible stimulus values. The readout <p></p><p></p><p><mi>s</mi><mo>̂</mo></p><p></p><p></p> can be viewed as an orthogonal projection of neural activities in the direction given by <b>a</b>. (C) Behavioral part of the model. The percept <p></p><p></p><p><mi>s</mi><mo>̂</mo></p><p></p><p></p> can be corrupted by decision noise <i>ξ</i>_<i>d</i>. Then it is thresholded to produce a binary choice <i>c</i>.<p>