Learning Rates Are Not All the Same: The Interpretation of Computational Model Parameters Depends on the Context

Reinforcement Learning (RL) has revolutionized the cognitive and brain sciences, explaining behavior from simple conditioning to problem solving, across the life span, and anchored in brain function. However, discrepancies in results are increasingly apparent between studies, particularly in the developmental literature. To better understand these, we investigated to which extent parameters generalize between tasks and models, and capture specific and uniquely interpretable (neuro)cognitive processes. 291 participants aged 8-30 years completed three learning tasks in a single session, and were fitted using state-of-the-art RL models. RL decision noise/exploration parameters generalized well between tasks, decreasing between ages 8-17. Learning rates for negative feedback did not generalize, and learning rates for positive feedback showed intermediate generalizability, dependent on task similarity. These findings can explain discrepancies in the existing literature. Future research therefore needs to carefully consider task characteristics when relating findings across studies, and develop strategies to computationally model how context impacts behavior.

Altered top-down and bottom-up processing of fear conditioning in panic disorder with agoraphobia

BackgroundAlthough several neurophysiological models have been proposed for panic disorder with agoraphobia (PD/AG), there is limited evidence from functional magnetic resonance imaging (fMRI) studies on key neural networks in PD/AG. Fear conditioning has been proposed to represent a central pathway for the development and maintenance of this disorder; however, its neural substrates remain elusive. The present study aimed to investigate the neural correlates of fear conditioning in PD/AG patients.MethodThe blood oxygen level-dependent (BOLD) response was measured using fMRI during a fear conditioning task. Indicators of differential conditioning, simple conditioning and safety signal processing were investigated in 60 PD/AG patients and 60 matched healthy controls.ResultsDifferential conditioning was associated with enhanced activation of the bilateral dorsal inferior frontal gyrus (IFG) whereas simple conditioning and safety signal processing were related to increased midbrain activation in PD/AG patients versus controls. Anxiety sensitivity was associated positively with the magnitude of midbrain activation.ConclusionsThe results suggest changes in top-down and bottom-up processes during fear conditioning in PD/AG that can be interpreted within a neural framework of defensive reactions mediating threat through distal (forebrain) versus proximal (midbrain) brain structures. Evidence is accumulating that this network plays a key role in the aetiopathogenesis of panic disorder.

Non-Myeloablative Allogeneic Hematopoietic Stem Cell Transplantation (allo-HSCT) for Patients with Severe Sickle Cell Disease (SCD)

Abstract 10 Allo-HSCT remains the only curative approach for patients with SCD, yet a high risk of procedural toxicities and graft-versus-host disease (GvHD) limit this approach in adult patients with end organ damage from SCD. We chose a low-dose radiation approach utilizing sirolimus based upon its unique ability to promote T cell tolerance. Entry criteria include irreversible SCD-related complications (stroke, nephropathy, or tricuspid regurgitant jet velocity >2.5 m/s), or reversible complications not ameliorated by a 6-month course of hydroxyurea (frequent vaso-occlusive crises or acute chest syndrome). Conditioning was achieved with 1mg/kg of alemtuzumab divided over 5 days, a single total body irradiation dose of 300cGy, and oral sirolimus targeting trough levels between 10–15 ng/ml. Donor chimerism was measured by microsatellite PCR among CD3 and CD14/15 positive white cells. Twenty-three patients have been transplanted to date, and their ages ranged from 17 to 64 years (median 28). All are alive at 2 months to 7 years post allo-HSCT. Conditioning was well-tolerated. All received unmanipulated G-CSF mobilized peripheral blood progenitors obtained from 8/8 HLA-matched siblings. CD34+ cell doses ranged from 5.5 to 31 × 10e6/kg (median 14.4), and CD3+ cell doses ranged from 1.6 to 5.4 × 10e8/kg (median 3.4). Three patients engrafted temporarily but lost their grafts between the 2nd and 3rd months post transplant and had recurrent SCD. Twenty patients engrafted with mean myeloid chimerism of 97.5% (median 89%) and CD3 chimerism of 42% (median 49%). Improvements in laboratory parameters of engrafted patients are as follows: pre vs post HSCT hemoglobin, 8.7 vs 12.1 g/dL; reticulocyte count, 181 vs 74 k/uL; LDH, 328 vs 202 units/L; total bilirubin, 2.87 vs 1.0 mg/dL. Hemoglobin electrophoresis revealed replacement by donor type hemoglobin in 19 patients by 1 year, with amelioration of the SCD phenotype allowing for therapeutic phlebotomy. In 17 patients at 1 year or more post allo-HSCT, 5 had CD3 chimerism >50% which allowed complete withdrawal of immunosuppression; they have maintained stable mixed chimerism. Two patients received preemptive treatment with ganciclovir or foscarnet for presumed CMV reactivation (blood PCR >300 genome/mL) before day 30. Treatment in both patients was discontinued 1 week later with negative blood PCR and prophylaxis with acyclovir was resumed. There were 2 patients with zoster at 2 and 3 years post allo-HSCT. Finally, no engrafted patient to date has developed any evidence of acute or chronic GVHD. Our results demonstrate that this relatively simple conditioning regimen is well-tolerated across a broad age range of adults with severe SCD, is effective in achieving stable mixed chimerism without the development of GVHD, and is sufficient to induce functional tolerance. Disclosures: No relevant conflicts of interest to declare.

Pavlov and Associationism

Pavlov's contribution to experimental psychology was to invent a technique that allowed him to undertake a prolonged and systematic series of well-controlled experiments that, astonishingly enough, uncovered many if not most of the phenomena of what is rightly called Pavlovian conditioning. It was not for another 30 years or more that English-speaking psychologists began to match that achievement. Of course there have been new developments and discoveries since his time. Two examples are discussed: the important role of variable associability or attention even in simple conditioning, and the rigorous application of associative learning theory to the behavior of adult humans.