Author Archives: carlos

► December 2015: Ben Scott and Christine Constantinople publish in eLife on noise properties in accumulation of evidence

ChristineConstantinople

Christine Constantinople

Ben Scott

Congratulations to Ben Scott and Christine Constantinople, co-first authors of a paper in eLife describing how noise variance in pulsatile accumulation of evidence tasks appears to not scale linearly with the number of pulses, but as the square of the number of pulses. This has a number of important implications for how the noise might be generated.

► February 2015: The lab will go en masse to COSYNE conference!

A remarkably lucky year for for us at COSYNE! Although their overall acceptance rate was 60%, all 7 of the abstracts we submitted to the COSYNE conference this year were accepted. (And yes, 0.6^7 is small — 0.028 😉 .)  Furthermore, Ann Duan’s abstract was chosen for a talk. Looks like the lab will have a blast at COSYNE this year.  🙂   Have fun skiing, folks!

 

► January 2015: Tim Hanks and Chuck Kopec publish in Nature on accumulation of evidence

Congratulations to postdocs Tim Hanks and Chuck Kopec, whose paper in Nature came out in advance online publication today. In this paper on the neural basis of decisions driven by accumulating evidence, they use our rat gradual accumulation of sensory evidence decision-making task (Brunton et al., Science, 2013), and they

  1. Show that rat cortical regions PPC and FOF have momentary-evidence-dependent ramping firing rates qualitatively very similar to analogous monkey cortices PPC and FEF. As in the monkey, these firing rate ramps are similar across the two regions.
  2. Develop a new method to estimate tuning curves (i.e., plots of firing rate as a function of a variable of interest) for accumulated evidence.
  3. Use their new method to show that despite the similarity in firing rate ramps, PPC and FOF have very distinct encodings of accumulating evidence: the PPC encodes the graded answer to “what is the value of the accumulating evidence”, whereas the FOF appears to have a more categorical encoding that can be roughly described as the categorical answer to “if the GO signal came now, which of the available decision outcomes should I choose?” This suggests that the FOF is more involved in response selection (choosing an available option, based on the accumulated evidence) than in gradual accumulation per se.
  4. Use halorhodopsin (eNpHR3.0) to test competing predictions: if the FOF is involved in the gradual accumulation process, which occurs throughout sensory evidence accumulation, then perturbing it at any point during the sensory stimulus should affect behavior. In contrast, if the FOF is primarily involved in response selection, perturbing it should have an effect on the behavior only near the end of the sensory stimulus, which is when response selection will occur. The data support the latter interpretation.

Together, the optogenetic and electrophysiological data suggest that despite its ramping firing rates, the FOF is involved in response selection, not graded accumulation. We wonder whether the results could hold for monkey FEF as well, and hope someone will use the methods we developed here to find out.

The results clarify the particular contribution of the FOF to decisions driven by accumulation of evidence. We are excited about applying the methods developed here to other brain regions linked to such decisions, in an effort to elucidate the different contributions of different brain areas, and understand how the whole circuit fits together.