skip to main content

Attention:

The NSF Public Access Repository (NSF-PAR) system and access will be unavailable from 11:00 PM ET on Friday, July 12 until 2:00 AM ET on Saturday, July 13 due to maintenance. We apologize for the inconvenience.


Title: Neural Substrates Mediating the Utility of Instrumental Divergence
We assessed the neural substrates mediating a recently demonstrated preference for environments with high levels of instrumental divergence – a formal index of flexible operant control. Across choice scenarios, participants chose between gambling environments that differed in terms of both instrumental divergence and expected monetary pay-offs. Using model-based fMRI, we found that activity in the ventromedial prefrontal cortex scaled with a divergence-based measure of expected utility that reflected the value of both divergence and monetary reward. Implications for a neural common currency for information theoretic and economic variables are discussed.  more » « less
Award ID(s):
1654187
NSF-PAR ID:
10132584
Author(s) / Creator(s):
Date Published:
Journal Name:
Proceedings of the 41st Annual Conference of the Cognitive Science Society
Page Range / eLocation ID:
2475-2480
Format(s):
Medium: X
Sponsoring Org:
National Science Foundation
More Like this
  1. A large literature has demonstrated an abnormal sense of agency (SOA) in schizophrenic individuals. One limitation of such studies is that they focus exclusively on cognitive or perceptual judgments, thus failing to address affective aspects of SOA. In our recent work, we have used instrumental divergence – the distance between outcome probability distributions associated with available actions – as a formal measure of agency, demonstrating an influence of this novel decision variable on behavioral choice preferences and associated neural computations in neurotypical adults. Here, we show that the preference for high instrumental divergence (i.e., for high-agency environments) is significantly modulated by individual differences in positive and negative schizotypy dimensions. Implications for future assessments of clinical populations are discussed. 
    more » « less
  2. Abstract

    Do related populations that are separated by barriers predictably evolve differences from one another over time, or is such divergence idiosyncratic and unpredictable? We test these alternatives by investigating patterns of trait evolution for 54 sister pairs of Andean forest birds that live in similar environments on either side of the arid Marañón Gap, a strong dispersal barrier for humid montane species. We measured divergence in both sexual (song and plumage) and ecological (beak size and beak shape) traits. Sexual traits evolve in a clock-like fashion, with trait divergence positively correlated with genetic distance (r = 0.6–0.7). In contrast, divergence in ecological traits is uncorrelated or only loosely correlated with genetic distance (r = 0.0–0.3). Thus, for geographically isolated Andean montane forest birds that live in similar environments, divergence is predictable in sexual traits, but not for ecological traits. This means that sexual trait divergence occurs independently of adaptive ecological divergence within the mega-diverse tropical Andean avifauna. Last, we show that variation in genetic divergence across a biogeographic barrier is associated with traits that are proxies for species’ opportunities for dispersal (low elevation limit and elevational niche breadth), but not with traits that are proxies for species’ dispersal abilities (hand-wing index and foraging strata).

     
    more » « less
  3. Our goal is to learn control policies for robots that provably generalize well to novel environments given a dataset of example environments. The key technical idea behind our approach is to leverage tools from generalization theory in machine learning by exploiting a precise analogy (which we present in the form of a reduction) between generalization of control policies to novel environments and generalization of hypotheses in the supervised learning setting. In particular, we utilize the probably approximately correct (PAC)-Bayes framework, which allows us to obtain upper bounds that hold with high probability on the expected cost of (stochastic) control policies across novel environments. We propose policy learning algorithms that explicitly seek to minimize this upper bound. The corresponding optimization problem can be solved using convex optimization (relative entropy programming in particular) in the setting where we are optimizing over a finite policy space. In the more general setting of continuously parameterized policies (e.g., neural network policies), we minimize this upper bound using stochastic gradient descent. We present simulated results of our approach applied to learning (1) reactive obstacle avoidance policies and (2) neural network-based grasping policies. We also present hardware results for the Parrot Swing drone navigating through different obstacle environments. Our examples demonstrate the potential of our approach to provide strong generalization guarantees for robotic systems with continuous state and action spaces, complicated (e.g., nonlinear) dynamics, rich sensory inputs (e.g., depth images), and neural network-based policies.

     
    more » « less
  4. A loss function measures the discrepancy between the true values (observations) and their estimated fits, for a given instance of data. A loss function is said to be proper (unbiased, Fisher consistent) if the fits are defined over a unit simplex, and the minimizer of the expected loss is the true underlying probability of the data. Typical examples are the zero-one loss, the quadratic loss and the Bernoulli log-likelihood loss (log-loss). In this work we show that for binary classification problems, the divergence associated with smooth, proper and convex loss functions is bounded from above by the Kullback-Leibler (KL) divergence, up to a multiplicative normalization constant. It implies that by minimizing the log-loss (associated with the KL divergence), we minimize an upper bound to any choice of loss functions from this set. This property justifies the broad use of log-loss in regression, decision trees, deep neural networks and many other applications. In addition, we show that the KL divergence bounds from above any separable Bregman divergence that is convex in its second argument (up to a multiplicative normalization constant). This result introduces a new set of divergence inequalities, similar to the well-known Pinsker inequality. 
    more » « less
  5. Abstract

    Adaptive divergence in response to environmental clines are expected to be common in species occupying heterogeneous environments. Despite numerous advances in techniques appropriate for non‐model species, gene–environment association studies in elasmobranchs are still scarce. The bronze whaler or copper shark (Carcharhinus brachyurus) is a large coastal shark with a wide distribution and one of the most exploited elasmobranchs in southern Africa. Here, we assessed the distribution of neutral and adaptive genomic diversity inC. brachyurusacross a highly heterogeneous environment in southern Africa based on genome‐wide SNPs obtained through a restriction site‐associated DNA method (3RAD). A combination of differentiation‐based genome‐scan (outflank) and genotype–environment analyses (redundancy analysis, latent factor mixed models) identified a total of 234 differentiation‐based outlier and candidate SNPs associated with bioclimatic variables. Analysis of 26,299 putatively neutral SNPs revealed moderate and evenly distributed levels of genomic diversity across sites from the east coast of South Africa to Angola. Multivariate and clustering analyses demonstrated a high degree of gene flow with no significant population structuring among or within ocean basins. In contrast, the putatively adaptive SNPs demonstrated the presence of two clusters and deep divergence between Angola and all other individuals from Namibia and South Africa. These results provide evidence for adaptive divergence in response to a heterogeneous seascape in a large, mobile shark despite high levels of gene flow. These results are expected to inform management strategies and policy at the national and regional level for conservation ofC. brachyuruspopulations.

     
    more » « less