The Sense of Confidence during Probabilistic Learning: A Normative Account.
The Sense of Confidence during Probabilistic Learning: A Normative Account.
复制标题
DOI：
10.1371/journal.pcbi.1004305
复制
发表时间：
2015-06
影响因子：
4.3
通讯作者：
Dehaene S
中科院分区：
文献类型：
--
作者：
Meyniel F;Schlunegger D;Dehaene S
随机环境中的学习包括从有限数量的噪声数据中估计模型，因此本质上是不确定的。然而，许多经典模型将学习过程简化为参数估计的更新，并忽略了学习也经常伴随着变量“了解感”或置信度这一事实。因此，这些主观置信度估计的特征和起源在很大程度上仍然未知。在这里，我们研究在学习过程中，人类是否不仅推断出其环境的模型，而且还从他们的推断中获得了准确的信心。在我们的实验中，人类估计了变化环境中两种视觉或听觉刺激之间的转换概率，并报告了他们的平均估计值和他们对此报告的置信度。为了形式化这两种估计之间的联系并与规范参考相比评估其准确性，我们为我们的任务得出了最佳推理策略。我们的结果表明，受试者准确地追踪了他们的推论正确的可能性。学习和评估对所学知识的信心似乎是两种密切相关的能力，这表明它们源于单一的推理过程。我们证明人类表现与最佳概率推理的几个属性相匹配。特别是，主观信心受到环境不确定性的影响，无论是在第一级（鉴于推断的随机特征，刺激发生的不确定性）还是在第二级（由于这些随机特征的意外变化而导致的不确定性）。置信度也会随着稳定期内观察数量的增加而适当增加。我们的结果支持这样的观点，即人类对环境的抽象非感官参数的推论具有定量的信心。这种能力不能简化为简单的启发式，它似乎是学习过程的核心属性。学习常常伴随着一种“了解的感觉”，即对获取相关信息越来越有信心。在这里，我们将这种内省能力形式化，并评估其在面对环境变化时的准确性和灵活性，这些变化会迫使一个人的心理模型发生变化。我们评估了这样一个假设：大脑充当统计学家，不仅准确跟踪最可能的环境状态，而且还准确跟踪与其自身推论相关的不确定性。我们表明，连续观察中主观置信度的变化与执行最佳推理的理想观察者的数学模型紧密相关。我们的结果表明，在学习过程中，大脑不断跟踪自身的不确定性，并且主观信心可能来自学习过程本身。因此，我们的结果表明，主观信心虽然目前尚未得到充分探索，但可以提供关键数据来更好地理解学习。
Learning in a stochastic environment consists of estimating a model from a limited amount of noisy data, and is therefore inherently uncertain. However, many classical models reduce the learning process to the updating of parameter estimates and neglect the fact that learning is also frequently accompanied by a variable “feeling of knowing” or confidence. The characteristics and the origin of these subjective confidence estimates thus remain largely unknown. Here we investigate whether, during learning, humans not only infer a model of their environment, but also derive an accurate sense of confidence from their inferences. In our experiment, humans estimated the transition probabilities between two visual or auditory stimuli in a changing environment, and reported their mean estimate and their confidence in this report. To formalize the link between both kinds of estimate and assess their accuracy in comparison to a normative reference, we derive the optimal inference strategy for our task. Our results indicate that subjects accurately track the likelihood that their inferences are correct. Learning and estimating confidence in what has been learned appear to be two intimately related abilities, suggesting that they arise from a single inference process. We show that human performance matches several properties of the optimal probabilistic inference. In particular, subjective confidence is impacted by environmental uncertainty, both at the first level (uncertainty in stimulus occurrence given the inferred stochastic characteristics) and at the second level (uncertainty due to unexpected changes in these stochastic characteristics). Confidence also increases appropriately with the number of observations within stable periods. Our results support the idea that humans possess a quantitative sense of confidence in their inferences about abstract non-sensory parameters of the environment. This ability cannot be reduced to simple heuristics, it seems instead a core property of the learning process. Learning is often accompanied by a “feeling of knowing”, a growing sense of confidence in having acquired the relevant information. Here, we formalize this introspective ability, and we evaluate its accuracy and its flexibility in the face of environmental changes that impose a revision of one’s mental model. We evaluate the hypothesis that the brain acts as a statistician that accurately tracks not only the most likely state of the environment, but also the uncertainty associated with its own inferences. We show that subjective confidence ratings varied across successive observations in tight parallel with a mathematical model of an ideal observer performing the optimal inference. Our results suggest that, during learning, the brain constantly keeps track of its own uncertainty, and that subjective confidence may derive from the learning process itself. Our results therefore suggest that subjective confidence, although currently under-explored, could provide key data to better understand learning.
登录
查看更多内容
影响因子：
19.9
作者：
Fiser, Jozsef;Berkes, Pietro;Orban, Gergo;Lengyel, Mate
通讯作者：
Lengyel, Mate
影响因子：
3.4
作者：
KAHNEMAN, D;TVERSKY, A
通讯作者：
TVERSKY, A
影响因子：
5.4
作者：
Juslin, P;Olsson, H
通讯作者：
Olsson, H
影响因子：
6.1
作者：
KAHNEMAN, D;TVERSKY, A
通讯作者：
TVERSKY, A
影响因子：
64.8
作者：
Kepecs, Adam;Uchida, Naoshige;Mainen, Zachary F.
通讯作者：
Mainen, Zachary F.