AVector Model of Causal Meaning
Phillip Wolff
Sign up for access to the world's latest research
checkGet notified about relevant papers
checkSave papers to use in your research
checkJoin the discussion with peers
checkTrack your impact
Abstract
This paper proposes a new model of causal meaning, the Vector Model, which formalizes a model of causation based on Talmys notions of force dynamics (Wolff, Song, & Driscoll, 2002). In the Vector Model, the concepts of CAUSE, ENABLE and PREVENT are distinguished from one another in terms of force vectors, their resultant and the relationship of each force vector to a target vector. The predictions of the model were tested in two experiments in which participants saw realistic 3D-animations of an inflatable boat moving through a pool of water. The boats movements were completely determined by the force vectors entered into a physics simulator. Participants linguistic descriptions of the animations were closely matched by those predicted by the model given the same force vectors as those used to produce the animations. Our model may have implications for the semantics of causal verbs as well as the perception of causal events.
Related papers
UC Merced Proceedings of the Annual Meeting of the Cognitive Science Society Title A Vector Model of Causal Meaning Permalink Publication Date A Vector Model of Causal Meaning
This paper proposes a new model of causal meaning, the Vector Model, which formalizes a model of causation based on Talmyís notions of force dynamics (Wolff, Song, & Driscoll, 2002). In the Vector Model, the concepts of CAUSE, ENABLE and PREVENT are distinguished from one another in terms of force vectors, their resultant and the relationship of each force vector to a target vector. The predictions of the model were tested in two experiments in which participants saw realistic 3D-animations of an inflatable boat moving through a pool of water. The boatís movements were completely determined by the force vectors entered into a physics simulator. Participantsí linguistic descriptions of the animations were closely matched by those predicted by the model given the same force vectors as those used to produce the animations. Our model may have implications for the semantics of causal verbs as well as the perception of causal events.
Causal Representation 1 Running head : CAUSAL REPRESENTATION Representing Causation
2007
The dynamics model, which is based on Talmy’s (1988) theory of force dynamics, characterizes causation as a pattern of forces and a position vector. In contrast to counterfactual and probabilistic models, the dynamics model naturally distinguishes between different cause-related concepts and explains the induction of causal relationships from single observations. Support for the model is provided in experiments in which participants categorized 3D animations of realistically rendered objects with trajectories that were wholly determined by the force vectors entered into a physics simulator. Experiments 1-3 showed that causal judgments are based on several forces, not just one. Experiment 4 demonstrated that people compute the resultant of forces using a qualitative decision rule. Experiments 5 and 6 showed that a dynamics approach extends to the representation of social causation. Implications for the relationship between causation and time are discussed.
Representing causation
Journal of Experimental Psychology: General, 2007
The dynamics model, which is based on L. Talmy's (1988) theory of force dynamics, characterizes causation as a pattern of forces and a position vector. In contrast to counterfactual and probabilistic models, the dynamics model naturally distinguishes between different cause-related concepts and explains the induction of causal relationships from single observations. Support for the model is provided in experiments in which participants categorized 3-D animations of realistically rendered objects with trajectories that were wholly determined by the force vectors entered into a physics simulator. Experiments 1-3 showed that causal judgments are based on several forces, not just one. Experiment 4 demonstrated that people compute the resultant of forces using a qualitative decision rule. Experiments 5 and 6 showed that a dynamics approach extends to the representation of social causation. Implications for the relationship between causation and time are discussed.
Causal reasoning with forces
Causal composition allows people to generate new causal relations by combining existing causal knowledge. We introduce a new computational model of such reasoning, the force theory, which holds that people compose causal relations by simulating the processes that join forces in the world, and compare this theory with the mental model theory (Khemlani et al., 2014) and the causal model theory (Sloman et al., 2009), which explain causal composition on the basis of mental models and structural equations, respectively. In one experiment, the force theory was uniquely able to account for people's ability to compose causal relationships from complex animations of real-world events. In three additional experiments, the force theory did as well as or better than the other two theories in explaining the causal compositions people generated from linguistically presented causal relations. Implications for causal learning and the hierarchical structure of causal knowledge are discussed.
Models of causation and causal verbs
2002
Introduction Recent models of causation place the concept of CAUSE within a broader framework of concepts that includes the notions of LETTING, HINDERING, HELPING and PREVENTING. In each model, the related concepts are defined in terms of a small set of conceptual distinctions. This paper examines the relationship between the cognitive and linguistic systems. Specifically, we investigate whether these models of causation can capture the distinctions that underlie the semantics of causal verbs. There are a number of structures that can be used to express the notion of CAUSE in English and other languages, including causal connectives (e.g., because), prepositions (because of, thanks to) and lexical causatives, i.e., verbs that encode both the notions of CAUSE and RESULT, as in Peter broke the stick (i.e., caused the stick to break). Yet another means of expressing CAUSE is the use of a verb that includes the notion of CAUSE without specification of a particular RESULT, as in the verbs cause, let, help or prevent (Ammon 1980,
The role of activity in visual impressions of causality
Acta Psychologica, 2006
Phenomenal causality is an illusion built on an incomplete perception. It is an illusion because we can have visual impressions of causality when no interaction between objects is actually taking place. It is an illusion built on an incomplete perception because causality as we understand it neglects some factors involved in objective descriptions of interactions between objects in terms of the laws of mechanics. So, why don't we perceive object interactions in accordance with the laws of mechanics? I Wrst consider what kinds of things can and cannot be causes perceptually, arguing that active objects can be causes and non-moving objects cannot be. Then, I argue that causal understanding originates with what we have the most direct experience of, our own actions on objects, and extends out from this point of origin to other domains of causality by a form of schema matching the interpretation of stimulus input by matching to abstracted stored representations of experiences. Schema matching raises the possibility of many more kinds of phenomenal causality than have hitherto been considered, and I conclude by suggesting some possibilities.
Caused-Motion and Caused-Position: Syntactic Patterns and Semantic Networks
Studies in Language, 2021
According to Goldberg (1995), placement verbs (such as put) are instantiated in the Caused-Motion Construction. Rohde (2001), however, argued that placement verbs in fact occur in a different construction, which she names the Caused-Position Construction, whose semantic value is not 'cause to move' but rather 'cause to be positioned'. The present paper redefines and justifies the postulation of Caused-Position Construction. The Caused-position Construction is compatible with not only placement verbs but also a variety of other verbs, such as verbs of creation (write or build) or certain stative verbs (want or need), many of which also occur in the Locative Inversion construction. Further, a similar distinction between Caused-Motion and Caused-Position can be attested in Mandarin as well, which suggests that the distinction between two patterns of spatial cau-sation may not be idiosyncratically confined to the English language but motivated by the general patterns of human cognition.
Caused motion events as complex structures
2007
From numerous works dealing with causative predicates it follows that their decomposition poses numerous problems. Decompositional formulas of causative events employ the abstract predicate ‘cause’ taking two arguments, the cause and the result. Causative events are thus represented as chains with the cause on the one pole and the result on the other: the causer x acts on the causee y, inducing a change in y. The resulting change in y may be either a process or a state. The present paper will focus on caused motion predicates. It will demonstrate that (a) caused motion events represent complex structures that may involve more than two subevents; the subevents are not only interrelated, but also display a hierarchical ordering (b) in spite of being clearly discernible, the subevents do not have an autonomous status (c) the interaction between the verb’s specific causative structure and its lexico-semantic structure manifests itself at a syntactic level.
Perception of Mechanical and Qualitative Causality:Answers from Developmental Psychology, Psychophysics, and Philosophy
We commonly have a strong sense of causality as events unfold. We often experience one event as causing another. Both perceptual and cognitive processes have been proposed as the explanation for causal experience. However, neither philosophy nor psychology have been able to provide decisive arguments one way or another. Theorists claiming that the origin of causal representation is the perceptual system argue that " certain physical events give an immediate causal impression, and that one can 'see' an object act on another object, produce in it certain changes, and modify it in one way or another " (Michotte, 1946/1963, p.15). On this view there exists a perceptual mechanism that transforms visual sequences of events into a representation of cause. Michotte hypothesized that causal perception was specific to caused motions of objects. More recently Susanna Siegel (2010) argued visual experiences might represent causal relations of a much larger variety. I argue that only the causation of motion is represented by the perceptual system. The questions this essay aims to answer are (i) is there a perceptual mechanism capable of attributing a causal relation, and (ii) if such a mechanism exists what kind of causal experiences are the result of its operations. My discussion of these questions unfolds as follows. Before confronting the questions of this essay I will explicate a few key terms. Then, I will distinguish two senses of causation and two theses concerning the visual representation of causation. The first thesis is the narrow causal thesis , which holds that caused motion of objects is the only mode of causation represented in visual experience. The broad causal thesis is not specific to what kind of causation is represented in visual experience. Following this, I will lay out the proposals for explaining causal experience from Michotte and Siegel's argument for the broad causal thesis. I then turn to evidence that infants representing launching events involves more than representations of spatiotemporal properties as well as evidence suggesting that computations performed by the perceptual system are responsible for the attribution of causation.
Related papers
Bodily forces, actions and the semantics of verbs
How can we talk about what we see? This is basically a question of translation between two different types of informationthe visual and the linguistic. 2 Signals from the retina are processed in several stages as they spread through the brain and they are transformed into a visual representation that we do not fully understand, even though there has been much progress during the last decades. Similarly, language is represented on several levels by the brain. Somehow these types of representation must be intertranslatable since we effortlessly can talk about what we see, and conversely, when we hear somebody tell a story we immediately form a vivid inner image of the narrative (and we can, more or less successfully, transform this image to a drawing or a sketch).
Semantic constraints on the caused-motion construction
Within the field of Cognitive Linguistics, Goldberg's (1995) approach to Cons-truction Grammar offers a detailed analysis on the semantic constraints that affect the caused-motion construction. The present paper aims to revise one specific constraint which Goldberg tentatively states as follows: «if the action denoted by the verb implies an effect other than motion, then a path of motion cannot be specified» (1995:170). In doing so, we will study the behavior of some «contact-by-impact» verbs (i.e., slap, smack, whack, knock and hit) making use of the analytical tools provided by the Lexi-cal-Constructional Model (LCM), developed by Ruiz de Mendoza and Mairal (2007). KEY WORDS: caused-motion construction, Lexical-Constructional Model (LCM), se-mantic constraints, lexical-constructional subsumption, contact-by-impact verbs. RESUMEN En el campo de la Lingüística Cognitiva, Goldberg (1995), en su acercamiento a la Gramática de Construcciones ofrece un análisis detallado de las rest...
The Role of Language in Expressing Agentivity in Caused Motion Events: A Cross-Linguistic Investigation
Frontiers in Psychology, 2022
While understanding and expressing causal relations are universal aspects of human cognition, language users may differ in their capacity to perceive, interpret, and express events. One source of variation in descriptions of caused motion events is agentivity, which refers to the attribution of a result to the agent's action. Depending on the perspective taken, the same event may be described with agentive or non-agentive interpretations. Does language play a role in how people construe and express caused motion events? The present study investigated the use of agentive vs. non-agentive language by speakers of different languages (i.e., monolingual speakers of English and Korean, and Korean learners of English). All three groups described prototypical causal events similarly, using agentive language (active transitive sentences). However, when it came to non-prototypical causal events (where the agent was not shown in the scene), they diverged in their choice of language: English speakers favored agentive language (passive transitive sentences), whereas Korean speakers preferred non-agentive language (intransitive sentences). Korean learners of English patterned with Korean speakers, demonstrating L1 influence on their use of English. These findings highlight the effects of language on motion event construal.
Representing verbs with force vectors
Theoretical Linguistics, 2012
Visualizing Causal Semantics Using Animations
IEEE Transactions on Visualization and Computer Graphics, 2000
Michotte's theory of ampliation suggests that causal relationships are perceived by objects animated under appropriate spatiotemporal conditions. We extend the theory of ampliation and propose that the immediate perception of complex causal relations is also dependent on a set of structural and temporal rules. We designed animated representations, based on Michotte's rules, for showing complex causal relationships or causal semantics. In this paper we describe a set of animations for showing semantics such as causal amplification, causal strength, causal dampening, and causal multiplicity. In a two part study we compared the effectiveness of both the static and animated representations. The first study (N=44) asked participants to recall passages that were previously displayed using both types of representations. Participants were 8% more accurate in recalling causal semantics when they were presented using animations instead of static graphs. In the second study (N=112) we evaluated the intuitiveness of the representations. Our results showed that while users were as accurate with the static graphs as with the animations, they were 9% faster in matching the correct causal statements in the animated condition. Overall our results show that animated diagrams that are designed based on perceptual rules such as those proposed by Michotte have the potential to facilitate comprehension of complex causal relations.
Force interaction in the expression of causation
Causal meanings in verbs such as cause, enable and prevent have been analyzed as having two components that correspond to two interacting forces or tendencies: one associated with the agent and one with the patient (Talmy 2000; Wolff 2007). In this research we extend a force-dynamic analysis to a wider range of causal and quasi-causal expressions such as lead to, because, and after. The “structural causal pluralism hypothesis” (Copley & Wolff 2014) is not supported; instead force dynamics is shown to be relevant to expressions throughout syntactic structure. We find that the applicability of the classical force-interaction analysis depends on (i) whether an Agent/Causer is represented in the syntax, and (ii) what kind of causing entity is conceptually represented: either one that generates its own force or one whose force emerges from an interaction with a field in the sense of Copley & Harley (2015) (e.g., a gravitational field). The latter case, we propose, suggests a criterion for force individuation. This account allows us to identify several classes of causal expressions and to further map out the division of labor between the grammatical and conceptual levels.
Physical Causality of Action Verbs in Grounded Language Understanding
Proceedings of the 54th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), 2016
Linguistics studies have shown that action verbs often denote some Change of State (CoS) as the result of an action. However, the causality of action verbs and its potential connection with the physical world has not been systematically explored. To address this limitation, this paper presents a study on physical causality of action verbs and their implied changes in the physical world. We first conducted a crowdsourcing experiment and identified eighteen categories of physical causality for action verbs. For a subset of these categories, we then defined a set of detectors that detect the corresponding change from visual perception of the physical environment. We further incorporated physical causality modeling and state detection in grounded language understanding. Our empirical studies have demonstrated the effectiveness of causality modeling in grounding language to perception.
The Elements of Cause and Effect
Cognitive science has developed many different theoretical approaches to causality. All of these approaches assume (sometimes implicitly) that we conceptualize causality in terms of cause and effect. It is argued that the elements-the conceptual primitives-of a cause and of an effect have not been identified yet. Thus, this paper sets out to carry out cognitive-linguistic analysis (systematic sentence manipulations) to answer the following two questions: (1) What are the mental elements that make up a cause? And what are the mental elements that make up an effect? It is argued that Talmy's force dynamics-with three revisions-can be used as a basic framework to identify these elements. The causal concepts investigated are: successful causation (CAUSE), failed causation (DESPITE), negative causation (PREVENT), and disengaged potential negative causation (ENABLE). This account of "force-dynamic elementary causality" is then also compared with other variants of force-dynamic theory. It is furthermore demonstrated how forcedynamic elementary causation can be integrated with epistemic and with counterfactual and probabilistic accounts. Finally, it is discussed how these causal elements might manifest in mental spatiotemporal structure.
The metaphoric motivation of the caused-motion construction: A case study of perception
International Journal of English Studies, 2013
This article addresses the caused-motion construction from the theoretical perspective of the Lexical Constructional Model (LCM). Within the LCM, the way in which lexical templates fuse with constructional templates is coerced by internal and external constraints. Internal constraints specify the conditions under which allow predicates to take part in a construction. External constraints take the form of high-level metaphoric and metonymic operations that affect lexical-constructional subsumption. This proposal makes use of the theoretical tools of the LCM with a view to exploring instantiations of the construction with verbs of perception. Apart from internal constraints, high-level metaphor will be found to play a prominent role in the construal of the examples under scrutiny. The study will suffice to point out that the semantics of the caused-motion construction needs to be understood with reference to the underlying metaphoric mappings.
A causal model theory of the meaning of cause, enable, and prevent
Cognitive science, 2009
The verbs cause, enable, and prevent express beliefs about the way the world works. We offer a theory of their meaning in terms of the structure of those beliefs expressed using qualitative properties of causal models, a graphical framework for representing causal structure. We propose that these verbs refer to a causal model relevant to a discourse and that ''A causes B'' expresses the belief that the causal model includes a link from A to B. ''A enables ⁄ allows B'' entails that the model includes a link from A to B, that A represents a category of events necessary for B, and that an alternative cause of B exists. ''A prevents B'' entails that the model includes a link from A to B and that A reduces the likelihood of B. This theory is able to account for the results of four experiments as well as a variety of existing data on human reasoning.
Loading Preview
Sorry, preview is currently unavailable. You can download the paper by clicking the button above.