Prediction‐based processes appear to play an important role in language. Few studies, however, have sought to test the relationship within individuals between prediction learning and natural language processing. This paper builds upon existing statistical learning work using a novel paradigm for studying the on‐line learning of predictive dependencies. Within this paradigm, a new “prediction task” is introduced that provides a sensitive index of individual differences for developing probabilistic sequential expectations. Across three interrelated experiments, the prediction task (...) and results thereof are used to bridge knowledge of the empirical relation between statistical learning and language within the context of nonadjacency processing. We first chart the trajectory for learning nonadjacencies, documenting individual differences in prediction learning. Subsequent simple recurrent network simulations then closely capture human performance patterns in the new paradigm. Finally, individual differences in prediction performances are shown to strongly correlate with participants’ sentence processing of complex, long‐distance dependencies in natural language. (shrink)

It is a popular view thatpractical deliberation excludes foreknowledge of one's choice. Wolfgang Spohn and Isaac Levi have argued that not even a purely probabilistic self-predictionis available to thedeliberator, if one takes subjective probabilities to be conceptually linked to betting rates. It makes no sense to have a betting rate for an option, for one's willingness to bet on the option depends on the net gain from the bet, in combination with the option's antecedent utility, rather than on the offered (...) odds. And even apart from this consideration, assigning probabilities to the options among which one is choosing is futile since such probabilities could be of no possible use in choice. The paper subjects these arguments to critical examination and suggests that, appearances notwithstanding, practical deliberation need not crowd outself-prediction. (shrink)

Legal decision-support systems have the potential to improve access to justice, administrative efficiency, and judicial consistency, but broad adoption of such systems is contingent on development of technologies with low knowledge-engineering, validation, and maintenance costs. This paper describes two approaches to an important form of legal decision support—explainable outcome prediction—that obviate both annotation of an entire decision corpus and manual processing of new cases. The first approach, which uses an attention network for prediction and attention weights to highlight (...) salient case text, was shown to be capable of predicting decisions, but attention-weight-based text highlighting did not demonstrably improve human decision speed or accuracy in an evaluation with 61 human subjects. The second approach, termed semi-supervised case annotation for legal explanations, exploits structural and semantic regularities in case corpora to identify textual patterns that have both predictable relationships to case decisions and explanatory value. (shrink)

According to the so-called ‘deliberation crowds out prediction’ thesis, while deliberating about what you’ll do, you cannot rationally have credences for what you’ll do – you cannot rationally have option-credences. Versions of the thesis have been defended by authors such as Spohn, Levi, Gilboa, Price, Louise, and others. After registering a number of concerns about the thesis, I rehearse and rebut many of the main arguments for it, grouped according to their main themes: agency, vacuity, betting, and decision-theoretical considerations. (...) I go on to suggest many possible theoretical roles for option-credences. -/- I locate the debate about the thesis in a broader discussion: Are there rational credence gaps – propositions to which one cannot rationally assign credences? If there are, they spell trouble for various foundations of Bayesian epistemology, including the usual ratio formula for conditional probability, conditionalization, decision theory, and independence. According to the thesis, credence gaps are completely mundane; they arise every time someone rationally deliberates. But these foundations are safe from any threat here, I contend, since the thesis is false. Deliberation welcomes prediction. (shrink)

Climate models don’t give us probabilistic forecasts. To interpret their results, alternatively, as serious possibilities seems problematic inasmuch as climate models rely on contrary-to-fact assumptions: why should we consider their implications as possible if their assumptions are known to be false? The paper explores a way to address this possibilistic challenge. It introduces the concepts of a perfect and of an imperfect credible world, and discusses whether climate models can be interpreted as imperfect credible worlds. That would allow one to (...) use models for possibilistic prediction and salvage widespread scientific practice. (shrink)

COVID-19 has substantially affected our lives during 2020. Since its beginning, several epidemiological models have been developed to investigate the specific dynamics of the disease. Early COVID-19 epidemiological models were purely statistical, based on a curve-fitting approach, and did not include causal knowledge about the disease. Yet, these models had predictive capacity; thus they were used to ground important political decisions, in virtue of the understanding of the dynamics of the pandemic that they offered. This raises a philosophical question about (...) how purely statistical models can yield understanding, and if so, what the relationship between prediction and understanding in these models is. Drawing on the model that was developed by the Institute of Health Metrics and Evaluation, we argue that early epidemiological models yielded a modality of understanding that we call descriptive understanding, which contrasts with the so-called explanatory understanding which is assumed to be the main form of scientific understanding. We spell out the exact details of how descriptive understanding works, and efficiently yields understanding of the phenomena. Finally, we vindicate the necessity of studying other modalities of understanding that go beyond the conventionally assumed explanatory understanding. (shrink)

In a recent paper, Jakob Hohwy argues that the emerging predictive processing perspective on cognition requires us to explain cognitive functioning in purely internalistic and neurocentric terms. The purpose of the present paper is to challenge the view that PP entails a wholesale rejection of positions that are interested in the embodied, embedded, extended, or enactive dimensions of cognitive processes. I will argue that Hohwy’s argument from analogy, which forces an evidentiary boundary into the picture, lacks the argumentative resources to (...) make a convincing case for the conceptual necessity to interpret PP in solely internalistic terms. For this reason, I will reconsider the postulation and explanatory role of the evidentiary boundary. I will arrive at an account of prediction error minimization and its foundation on the free energy principle that is fully consistent with approaches to cognition that emphasize the embodied and interactive properties of cognitive processes. This gives rise to the suggestion that explanatory pluralism about the application of PP is to be preferred over Hohwy’s explanatory monism that follows from his internalistic and neurocentric view of predictive cognitive systems. (shrink)

Every model leaves out or distorts some factors that are causally connected to its target phenomenon -- the phenomenon that it seeks to predict or explain. If we want to make predictions, and we want to base decisions on those predictions, what is it safe to omit or to simplify, and what ought a causal model to describe fully and correctly? A schematic answer: the factors that matter are those that make a difference to the target phenomenon. There are several (...) ways to understand differencemaking. This paper advances a view as to which is the most relevant to the forecaster and the decision-maker. It turns out that the right notion of differencemaking for thinking about idealization in prediction is also the right notion for thinking about idealization in explanation; this suggests a carefully circumscribed version of Hempel's famous thesis that there is a symmetry between explanation and prediction. (shrink)

Perhaps because both explanation and prediction are key components to understanding, philosophers and psychologists often portray these two abilities as though they arise from the same competence, and sometimes they are taken to be the same competence. When explanation and prediction are associated in this way, they are taken to be two expressions of a single cognitive capacity that differ from one another only pragmatically. If the difference between prediction and explanation of human behavior is merely pragmatic, (...) then anytime I predict someone’s future behavior, I would at that moment also have an explanation of the behavior. I argue that advocates of both the theory theory and the simulation theory accept the symmetry of psychological prediction and explanation. However, there is very good reason to believe that this hypothesis is false. Just as we can predict the occurrence of some physical phenomena that we have no explanation for, we are also able to make accurate predictions of intentional behavior without having an explanation. Rather than requiring mental state attribution, I argue that the prediction of human behavior is most often accomplished by statistical induction rather than through an appeal to mental states. However, explanations are not given in these terms. (shrink)

From ‘connecting the dots’ and finding ‘the needle in the haystack’ to predictive policing and data mining for counterinsurgency, security professionals have increasingly adopted the language and methods of computing for the purposes of prediction. Digital devices and big data appear to offer answers to a wide array of problems of (in)security by promising insights into unknown futures. This article investigates the transformation of prediction today by placing it within governmental apparatuses of discipline, biopower and big data. Unlike (...) disciplinary and biopolitical governmentality, we argue that prediction with big data is underpinned by the production of a different time/space of ‘between-ness’. The digital mode of prediction with big data reconfigures how we are governed today, which we illustrate through an analysis of how predictive policing actualizes between-ness as hotspots and near-real-time decisions. (shrink)

This book develops a philosophico-methodological analysis of prediction and its role in economics. Prediction plays a key role in economics in various ways. It can be seen as a basic science, as an applied science, and in the application of this science. First, it is used by economic theory in order to test the available knowledge. In this regard, prediction has been presented as the scientific test for economics as a science. Second, prediction provides a content (...) regarding the possible future that can be used for prescription in applied economics. Thus, it can be used as a guide for economic policy, id est, as knowledge concerning the future to be employed for the resolution of specific problems. Third, prediction also has a role in the application of this science in the public arena. This is through the decision-making of the agents - individuals or organizations - in quite different settings, both in the realm of microeconomics and macroeconomics. Within this context, the research is organized in five parts, which discuss relevant aspects of the role of prediction in economics: I) The problem of prediction as a test for a science; II) The general orientation in methodology of science and the problem of prediction as a scientific test; III) The methodological framework of social sciences and economics: Incidence for prediction as a test; IV) Epistemology and methodology of economic prediction: Rationality and empirical approaches; and V) Methodological aspects of economic prediction: From description to prescription. Thus, the book is of interest for philosophers and economists as well as policy-makers seeking to ascertain the roots of their performance. The style used lends itself to a wide audience.. (shrink)

The Recursive Bayesian Net (RBN) formalism was originally developed for modelling nested causal relationships. In this paper we argue that the formalism can also be applied to modelling the hierarchical structure of mechanisms. The resulting network contains quantitative information about probabilities, as well as qualitative information about mechanistic structure and causal relations. Since information about probabilities, mechanisms and causal relations is vital for prediction, explanation and control respectively, an RBN can be applied to all these tasks. We show in (...) particular how a simple two-level RBN can be used to model a mechanism in cancer science. The higher level of our model contains variables at the clinical level, while the lower level maps the structure of the cell's mechanism for apoptosis. (shrink)

For multiple reasons, deliberating groups often converge on falsehood rather than truth. Individual errors may be amplified rather than cured. Group members may fall victim to a bad cascade, either informational or reputational. Deliberators may emphasize shared information at the expense of uniquely held information. Finally, group polarization may lead even rational people to unjustified extremism. By contrast, prediction markets often produce accurate results, because they create strong incentives for revelation of privately held knowledge and succeed in aggregating widely (...) dispersed information. The success of prediction markets offers a set of lessons for increasing the likelihood that groups can obtain the information that their members have. (shrink)

In this paper I reconstruct and critically examine the reasoning leading to the famous prediction of the ‘omega minus’ particle by M. Gell-Mann and Y. Ne’eman (in 1962) on the basis of a symmetry classification scheme. While the peculiarity of this prediction has occasionally been noticed in the literature, a detailed treatment of the methodological problems it poses has not been offered yet. By spelling out the characteristics of this type of prediction, I aim to underscore the (...) challenges raised by this episode to standard scientific methodology, especially to the traditional deductive-nomological account of prediction. (shrink)

This paper argues against the deductive reconstruction of scientific prediction, that is, against the view that in prediction the predicted event follows deductively from the laws and initial conditions that are the basis of the prediction. The major argument of the paper is intended to show that the deductive reconstruction is an inaccurate reconstruction of actual scientific procedure. Our reason for maintaining that it is inaccurate is that if the deductive reconstruction were an accurate reconstruction, then scientific (...) prediction would be impossible. (shrink)

It is widely believed that if a piece of evidence for a theory was known at the time the theory was proposed, then it does not confirm the theory as strongly as it would if the evidence had been discovered after the theory was proposed. I shall call this view the predictivist thesis. Those who have endorsed it include Leibniz (1678), Huygens (1690, preface), Whewell (1847 vol. 2, p. 64f.), Peirce (1883), Duhem (1914, ch. II, §5), Popper (1965, p. 241f.), (...) Lakatos (1970, p. 123), and Kuhn (1977, p. 322). On the other hand, the thesis has been rejected by a number of philosophers, including Mill (1872 bk. III, ch. 14, §6), Keynes (1921, p. 305), Rosenkrantz (1977, p. 169f.), Horwich (1982, pp. 108-117) and Schlesinger (1987). Others, while not rejecting the predictivist thesis, nevertheless regard the justification of the thesis as problematic; these include Hempel (1966, p. 38) and Gardner (1982). (shrink)

In the past two decades, reinforcement learning has become a popular framework for understanding brain function. A key component of RL models, prediction error, has been associated with neural signals throughout the brain, including subcortical nuclei, primary sensory cortices, and prefrontal cortex. Depending on the location in which activity is observed, the functional interpretation of prediction error may change: Prediction errors may reflect a discrepancy in the anticipated and actual value of reward, a signal indicating the salience (...) or novelty of a stimulus, and many other interpretations. Anterior cingulate cortex has long been recognized as a region involved in processing behavioral error, and recent computational models of the region have expanded this interpretation to include a more general role for the region in predicting likely events, broadly construed, and signaling deviations between expected and observed events. Ongoing modeling work investigating the interaction between ACC and additional regions involved in cognitive control suggests an even broader role for cingulate in computing a hierarchically structured surprise signal critical for learning models of the environment. The result is a predictive coding model of the frontal lobes, suggesting that predictive coding may be a unifying computational principle across the neocortex. (shrink)

The universal optimality theorem for metainduction works for epistemic agents faced with a choice among finitely many prediction methods. Eckhart Arnold and Tom Sterkenburg objected that it breaks...

Recent studies indicate that the processing of an unexpected word is costly when the initial, disconfirmed prediction was strong. This penalty was suggested to stem from commitment to the strongly predicted word, requiring its inhibition when disconfirmed. Additional studies show that comprehenders rationally adapt their predictions in different situations. In the current study, we hypothesized that since the disconfirmation of strong predictions incurs costs, it would also trigger adaptation mechanisms influencing the processing of subsequent strong predictions. In two experiments, (...) participants made speeded congruency judgments on two-word phrases in which the first word was either highly constraining or not. We manipulated the proportion of disconfirmed predictions in highly constraining contexts between participants. The results provide additional evidence of the costs associated with the disconfirmation of strong predictions. Moreover, they show a reduction in these costs when participants experience a high proportion of disconfirmed strong predictions throughout the experiment, indicating that participants adjust the strength of their predictions when strong prediction is discouraged. We formulate a Bayesian adaptation model whereby prediction failure cost is weighted by the participant’s belief about the likelihood of encountering the expected word, and show that it accounts for the trial-by-trial data. (shrink)

There is considerable disagreement about the epistemic value of novel predictive success, i.e. when a scientist predicts an unexpected phenomenon, experiments are conducted, and the prediction proves to be accurate. We survey the field on this question, noting both fully articulated views such as weak and strong predictivism, and more nascent views, such as pluralist reasons for the instrumental value of prediction. By examining the various reasons offered for the value of prediction across a range of inferential (...) contexts , we can see that neither weak nor strong predictivism captures all of the reasons for valuing prediction available. A third path is presented, Pluralist Instrumental Predictivism; PIP for short. (shrink)

First published in 1949 expressly to introduce logical positivism to English speakers. Reichenbach, with Rudolph Carnap, founded logical positivism, a form of epistemofogy that privileged scientific over metaphysical truths.

Evidence for a hypothesis typically confirms the hypothesis more if the evidence was predicted than if it was accommodated. Or so I argued in previous papers, where I also developed an analysis of why this should be so. But this was all a mistake if Howson and Franklin (1991) are to be believed. In this paper, I show why they are not to be believed. I also identify a grain of truth that may have been dimly grasped by those Bayesians (...) who deny the confirmatory value of prediction. (shrink)

People tend to perceive ongoing continuous activity as series of discrete events. This partitioning of continuous activity may occur, in part, because events correspond to dynamic patterns that have recurred across different contexts. Recurring patterns may lead to reliable sequential dependencies in observers' experiences, which then can be used to guide perception. The current set of simulations investigated whether this statistical structure within events can be used 1) to develop stable internal representations that facilitate perception and 2) to learn when (...) to update such representations in a self-organizing manner. These simulations demonstrate that experience with recurring patterns enables a system to accurately predict upcoming stimuli within an event, to identify boundaries between such events based on transient increases in prediction error, and to use such boundaries to improve prediction about subsequent activities. (shrink)

The disambiguation of a syntactically ambiguous sentence in favor of a less preferred parse can lead to slower reading at the disambiguation point. This phenomenon, referred to as a garden‐path effect, has motivated models in which readers initially maintain only a subset of the possible parses of the sentence, and subsequently require time‐consuming reanalysis to reconstruct a discarded parse. A more recent proposal argues that the garden‐path effect can be reduced to surprisal arising in a fully parallel parser: words consistent (...) with the initially dispreferred but ultimately correct parse are simply less predictable than those consistent with the incorrect parse. Since predictability has pervasive effects in reading far beyond garden‐path sentences, this account, which dispenses with reanalysis mechanisms, is more parsimonious. Crucially, it predicts a linear effect of surprisal: the garden‐path effect is expected to be proportional to the difference in word surprisal between the ultimately correct and ultimately incorrect interpretations. To test this prediction, we used recurrent neural network language models to estimate word‐by‐word surprisal for three temporarily ambiguous constructions. We then estimated the slowdown attributed to each bit of surprisal from human self‐paced reading times, and used that quantity to predict syntactic disambiguation difficulty. Surprisal successfully predicted the existence of garden‐path effects, but drastically underpredicted their magnitude, and failed to predict their relative severity across constructions. We conclude that a full explanation of syntactic disambiguation difficulty may require recovery mechanisms beyond predictability. (shrink)

Hans Reichenbach was a formidable figure in early-twentieth-century philosophy of science. Educated in Germany, he was influential in establishing the so-called Berlin Circle, a companion group to the Vienna Circle founded by his colleague Rudolph Carnap. The movement they founded—usually known as "logical positivism," although it is more precisely known as "scientific philosophy" or "logical empiricism"—was a form of epistemology that privileged scientific over metaphysical truths. Reichenbach, like other young philosophers of the exact sciences of his generation, was deeply impressed (...) by the far-reaching changes in physics brought about by Einstein's special and general theories of relativity. Reichenbach responded to scientific advances by doing fundamental work in space-time theories, in quantum mechanics, in statistical mechanics, and in the development of probability theory—making him the most important philosopher of physics in the first generation of logical empiricism. Forced from his academic position by the Nazi race laws in 1933, Reichenbach wrote _Experience and Prediction_ at the University of Istanbul, where had had fled, expressly to introduce logical positivism to English speakers. In the two decades following World War II, during the explosion of scientific advances in North America, logical positivism was the reigning theory of the philosophy of science and Reichenbach was at the peak of his career. But, inevitably, support for logical positivism began to wane as it became obvious that the justification of scientific theories could not be entirely resolved by relying on strictly formal, technical processes. The growth of the discipline of the history of philosophy of science, which has created an audience of scholars eager for seminal classics in scientific philosophy, and the evidence supporting a historicist paradigm within logical positivism are two important reasons to make _Experience and Prediction_ available once again. "Hans Reichenbach's_ Experience and Prediction_ is one of the most important books in twentieth-century philosophy of science. Its author was, along with Rudolf Carnap, one of the two principal ambassadors to North America of the exciting new European philosophical movement known here under the names 'Logical Positivism' and 'Scientific Philosophy.' In 1938, when the book was published, Reichenbach was an exile from his native Germany, teaching in Istanbul, Turkey, and about to emigrate to the United States to take up a prestigious position at UCLA. He wrote_ Experience and Prediction_ in English as his calling card to his new American colleagues. More than any other single book,_ Experience and Prediction_ set the agenda for the new discipline of the philosophy of science that was to emerge after World War II as, perhaps, the most exciting new area in North American philosophy. Many of the problems still at the focus of discussion were given their classic formulations in this book. Long out of print,_ Experience and Prediction_ appears here in a new edition accompanied by a splendid historical introduction by the noted young philosopher and historian of the philosophy of science, Alan Richardson. A jewel of a book may once again be appreciated in its proper setting." —Don A. Howard, University of Notre Dame. (shrink)

Here we will discuss the necessary relationship between both prediction and human behavior and meaning and human behavior. The main assumption upon which our thesis rests is that knowing is for the sake of acting and that, consequently, the symbolic process is continuous with overt bodily behavior and with the environment of the knower. A corollary to this assumption is: the locus of meaning is in behavior. Possibly after reading the article it will be clear that meanings presuppose conduct (...) of which we can take cognizance, and that there is a difference between behaving and taking cognizance of behavior. And although the locus of meaning is in behavior, meanings emerge only in relation to the symbolic process. In a sense we suffer, endure, live through behavior, but in the bare suffering of it there may be no meaning, for meaning requires also the symbolic process. (shrink)

The goal of perceptual systems is to allow organisms to adaptively respond to ecologically relevant stimuli. Because all perceptual inputs are ambiguous, perception needs to rely on prior knowledge accumulated over evolutionary and developmental time to turn sensory energy into information useful for guiding behavior. It remains controversial whether the guidance of perception extends to cognitive states or is locked up in a “cognitively impenetrable” part of perception. I argue that expectations, knowledge, and task demands can shape perception at multiple (...) levels, leaving no part untouched. The position advocated here is broadly consistent with the notion that perceptual systems strive to minimize prediction error en route to globally optimal solutions :181–204, 2013). On this view, penetrability should be expected whenever constraining lower-level processes by higher level knowledge is minimizes global prediction error. Just as Fodor feared cognitive penetration of perception threatens theory-neutral observation and the distinction between observation and inference. However, because theories themselves are constrained by the task of minimizing prediction error, theory-laden observation turns out to be superior to theory-free observation in turning sensory energy into useful information. (shrink)

Jonathan Walkan challenges cognitive science's dominant model of mental representation and proposes a novel, well-devised alternative. The traditional view in the cognitive sciences uses a linguistic model of mental representation. That logic-based model of cognition informs and constrains both the classical tradition of artificial intelligence and modeling in the connectionist tradition. It falls short, however, when confronted by the frame problem---the lack of a principled way to determine which features of a representation must be updated when new information becomes available. (...) So far, proposed alternatives, including the imagistic model, have not resolved the problem. Waskan proposes the Intrinsic Cognitive Models hypothesis, according to which representational states can be concpetualized as the cognitive equivalent of scale models.Waskan argues further that the proposal that humans harbor and manipulate cognitive counterparts to scale models offers the only viable explanation for what most clearly differentiates humans from other creatures: the capacity to engage in truth-preserving manipulation of representations. The ICM hypothesis, he claims, can be distinguished from sentence-based accounts of truth preservation in a way that is fully compatible with what is known about the brain. (shrink)