The Elements of Cause and Effect
Transcript of The Elements of Cause and Effect
The elements of cause and effect
Abstract
Cognitive science has developed many different theoretical approaches to causal-
ity. All of these approaches assume (sometimes implicitly) that we conceptualize
causality in terms of cause and effect. It is argued that the elements—the concep-
tual 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 el-
ements that make up a cause? And (2) 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 investi-
gated 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 force-
dynamic elementary causation can be integrated with epistemic and with coun-
terfactual and probabilistic accounts. Finally, it is discussed how these causal
elements might manifest in mental spatiotemporal structure.
Keywords: force dynamics; causation; counterfactual; Talmyan concept struc-
turing; mental time line
(about 13800 words, including references)
1
Contents
1 Introduction 3
2 Revising three basic concepts in Talmy’s force dynamics 4
2.1 Revising Talmy’s force-dynamic “action/rest” dichotomy . . . . . . 4
2.2 Revising Talmy’s force-dynamic “tendency” . . . . . . . . . . . . . 9
2.3 Interpreting Talmy’s “onset/extended causation” distinction in terms
of cause and effect . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
3 Successful causation (s-causation) 15
4 Failed causation (f-causation) 19
5 Negative causation (n-causation) 23
6 Disengaged potential negative causation (dpn-causation) 29
7 Discussion 34
7.1 Digital and analog representation in mental causality . . . . . . . . 36
7.2 Basic opposition vs. basic opposition/concordance . . . . . . . . . . 37
7.3 Force-dynamic elementary causation and epistemics . . . . . . . . . 39
7.4 Force-dynamic elementary causation and counterfactual and prob-
abilistic theories of causation . . . . . . . . . . . . . . . . . . . . . 41
7.5 Force-dynamic elementary causation and the mental time line . . . 44
(table of contents is only added for overview purposes; will be removed in the final
version)
2
1 Introduction
Cognitive science has developed many different theoretical approaches to causality.
Among the most prominent contemporary ones are: force dynamics (Barbey & Wolff
2007; Jackendoff 1990, 1996; Talmy 1985, 1988, 2000a; Wolff & Song 2003; Wolff 2003,
2007), causal model theory (Ali et al. 2011; Sloman 2005; Sloman et al. 2009), mental
model theory (Goldvarg & Johnson-Laird 2001), and counterfactual and probabilistic
accounts of causality (e.g., Pearl 2000; Spirtes et al. 2000). All of these approaches
assume, either explicitly or implicitly, that we conceptualize causality in terms of cause
and effect. This basic assumption is also in accordance with classical philosophical
treatments of causality (e.g. Hume 1748/1999; Kant 1787/2009). Yet despite this wide
agreement on the mental existence of cause and effect, this article argues that the
elements—the conceptual primitives—of a cause and of an effect have not been iden-
tified yet. In the light of this, this article sets out to carry out cognitive-linguistic
analysis (systematic sentence manipulations) to answer the two following questions:
(1) What are the mental elements that make up a cause? And (2) what are the mental
elements that make up an effect? It is argued that Talmy’s force dynamics (Talmy
1985, 1988, 2000a)—with three revisions (Section 2)—can be used as a basis to iden-
tify these elements. These three revisions of Talmy’s force dynamics are necessary for
two main reasons. First, Talmy’s account cannot handle all cases of physical causation
(Section 2.1). Secondly, while Talmy has dealt with some metaphorical extensions of
force dynamics, for certain causation phenomena that go beyond the physical domain
it is still unclear how Talmy’s account could be applied —for instance for causation in-
volving ideational entities (cf. Sections 3–6). Thus while the present account of mental
causality1 is still largely in spirit with Talmy’s original account, it additionally attempts1The term “mental causality” is ambiguous. On the one hand, it can refer to how we mentally
construe causality. This is the meaning intended in this article. On the other hand, “mental causality”can also mean that thinking is subject to some general causal laws. This latter possibility is not the
3
to provide an account that can in principle be applied to all causal conceptualization—
to all mentally construed physical causation and to all mentally construed causation
that goes beyond the physical domain. The causal concepts investigated are: success-
ful causation (CAUSE) (Section 3), failed causation (DESPITE) (Section 4), negative
causation (PREVENT) (Section 5), and what this article characterizes as “disengaged
potential negative causation” (ENABLE) (Section 6). The so-developed account is then
discussed in relation to: other variants of force-dynamic theory (Sections 7.1–7.2); epis-
temic phenomena (Section 7.3); counterfactual and probabilistic reasoning accounts of
causality (Section 7.4); and spatiotemporal structure (Section 7.5).
2 Revising three basic concepts in Talmy’s force dy-
namics
In this section, reasons are given why three basic concepts of Talmy’s force dynamics
need to be revised or refined to some degree: the “action/rest” dichotomy (Section 2.1),
the notion of a force-dynamic “tendency” (Section 2.2), and the “onset/extended cau-
sation” distinction (Section 2.3).
2.1 Revising Talmy’s force-dynamic “action/rest” dichotomy
Talmy’s force dynamics (Talmy 1985, 1988; 2000a; 2011, pp. 633–634) involves the
assumption that mental causality involves the conceptualization of entities that are in
opposition and that are engaged in force interaction, a view that can be traced back
to Kant (1787/2009, p. 229; 1781/1998, p. 314).2 Consider an example adapted from
subject of the current investigation.2Jackendoff (1990) and Wolff & Song (2003) challenge Talmy’s assumption that most force-dynamic
interactions involve opposition and propose that many of these interactions also basically involveconcordance. This issue is taken up in Section 7.2.
4
Talmy (2000a, p. 416):
(1) The ball kept rolling because of the wind.3
Using Talmy’s force-dynamic framework, in (1) the ball takes on the role of what is
called the agonist (Ago). In Talmy’s framework, Ago is an entity with an intrinsic force
tendency either toward action or toward rest. In (1) Ago (the ball) is conceptualized as
having the intrinsic tendency toward rest. In Talmy’s force dynamics, there are always
two force entities. The second force entity is called the antagonist (Ant). In (1) this role
is taken on by the wind. Ant is always conceptualized as (i) having the opposite intrinsic
force tendency of Ago and as (ii) as trying to impose this opposite tendency onto Ago.
Thus in Talmy’s force dynamics if Ago has the intrinsic tendency toward rest, Ant will
attempt to make it act and if Ago has the intrinsic tendency toward action, Ant will
attempt to make it stop. Consequently in (1) the wind is conceptualized as having the
opposite intrinsic force tendency of the ball: it is conceptualized as having the tendency
toward motion which it tries to impose onto the ball. The resultant of the interaction
always relates to Ant. This resultant depends on which intrinsic force is conceptualized
as being stronger. In conceptualizations underlying the use of because—as in (1)—Ant
is always stronger. Thus, in (1) Ant’s tendency (the wind’s intrinsic tendency toward
motion) is stronger than Ago’s tendency (the ball’s intrinsic tendency to stationariness)
and hence the ball is conceptualized as being moved by the wind. (It could of course
also be that Ago’s—here the ball’s—force is stronger; see Section 4.)3The generated example sentences that start here have been approved as a set by one native
American English speaker.
5
Figure 1: Talmy’s system of force-dynamic diagramming (Talmy 2000a, p. 414).
Talmy uses a specific system of diagramming (Figure 1) to represent force-dynamic
interactions. Thus, (1) would be notated as shown in Figure 2.
Figure 2: Talmy’s system of force-dynamic diagramming for the conceptual structureunderlying because (as well as other forms of s-causation, see below).
Thus in Figure 2, using (1) as an example, Ago (the ball) with its intrinsic tendency
for rest is diagrammed as a circle (=Ago) with a dot (= rest). Ant (the wind) is
diagrammed as a concave figure. Ant being stronger (+) causes the ball to move (>, on
the resultant line). I shall adopt a notational variant of this diagramming system that, I
6
think, will help to flesh out the cause-and-effect relationship that is implicitly assumed
in Talmy’s force dynamics. With this variant, (1) can be notated in the following form:
(2) Notational variant of Talmy’s system of force-dynamic diagramming, exemplified
with the conceptual structure underlying because
C: Ago-T (–), Ant-T>(+) → E: Ago>
The left side of the arrow (→) now represents the cause (C) and the right side the
effect (E). Thus the cause (C) can be identified as involving the following scenario:
Ago has the tendency (Ago-T) for rest ( ), Ant imposes its stronger (+) tendency
(Ant-T) of motion action (>) onto the weaker (–) Ago. This intervention of Ant on
Ago has the effect (E) that Ago is forced to move (Ago>).4
As has just been shown, in Talmy’s force dynamics (Talmy 1985, 1988, 2000a),
when Ago’s intrinsic tendency is rest, then Ant’s intrinsic tendency must be action
and vice versa. However, Talmy’s rest/action dichotomy cannot essentially capture
the cause-and-effect relationship that is expressed by because, as the following example
demonstrates:
(3) The rolling ball changed direction because of the wind.
Clearly, in (3) the ball (Ago) has already at the outset the intrinsic tendency to ac-
tion (movement) and the stronger wind (Ant) changes this intrinsic tendency to yet
another motion action. In other words, one cannot identify a toward rest/toward action
dichotomy, which we should be able to in Talmy’s force-dynamic framework. Thus
Talmy’s action/rest dichotomy cannot essentially capture the relationship of cause and
effect (or the conceptual structure underlying because). A solution to this problem is
to abstract away from the action/rest dichotomy in the following way:4On the key importance of the concept of “intervention” in mental causality see Sloman (2005);
cf. also Section 7.4).
7
(4) Redefinition of Talmy’s force-dynamic action/rest dichotomy
a. Ago (agonist): a force-dynamic entity with any given intrinsic tendency
(tendency x)
b. Ant (antagonist): a force dynamic entity that (i) has an intrinsic tendency
which is different (diff) to the intrinsic tendency of Ago (tendency xdiff) and
that (ii) attempts to impose xdiff onto Ago
(4a) proposes to replace Talmy’s notion of an agonist which always has a tendency either
toward action or rest with the more abstract notion that Ago can in principle have
any given intrinsic tendency. This “any-ness” is termed x, hence the term tendency x.
(4b) proposes to replace Talmy’s notion of an antagonist which always has an opposite
tendency that is either action or rest with the more abstract notion that Ant in principle
always has an intrinsic tendency that is simply different to the intrinsic tendency of
Ago (I will notate this as tendency xdiff).5 Overall, (4) leads to the following revision
of (2):
(5) C: Ago-Tx(–), Ant-Txdiff(+) → E: Ago-xdiff
The cause (C) can now be identified as involving the following scenario: Ago has any
tendency (Tx) and Ant then imposes its stronger different tendency (Txdiff) onto the
weaker (–) Ago. This intervention of Ant has the effect (E) that Ago takes on the
imposed different value of Ant (Ago-xdiff). Thus, (5), unlike Talmy’s (2), can capture (3):
The ball has a given tendency x (here, rolling in a certain direction), then there is Ant’s5When presented with example (3), Talmy (2012, personal communication) agreed that his ac-
tion/rest dichotomy cannot capture it. I then suggested that Ago can have any given tendency (ten-dency x) and that Ant could have any tendency that could be construed as a rough opposite (I calledit an intrinsic tendency that is “oppositable” to the intrinsic tendency of Ago). It was then Talmy whoin this discussion immediately came up with a much simpler solution for Ant: that Ant’s tendency issimply different to Ago’s tendency. I do not know if Talmy would stick to his insight after reflectingabout it. But, as far as I can see, replacing the action/rest dichotomy with an x/xdiff dichotomy is thecorrect solution—a solution that works for all instances of s-causation (see Section 3) and f-causation(see Section 4) that I can think of.
8
intervention which imposes Ant’s stronger different tendency (its tendency of making
the ball roll in another direction) on Ago—with the effect (E) that Ago is forced to
roll on in the different direction given by the wind. As mentioned above, the notation
developed in this article—for instance, as in (5)—is not meant to suggest a different
mental reality than Talmy’s system of force-dynamic diagramming (Talmy, 1976, p. 62;
1985; 1988; 2000a, p. 414). As also mentioned, the main function of the notational
system is to separate the force-dynamic cause-and-effect elements—cause: elements on
left side of arrow; effect: elements on right side of arrow.
2.2 Revising Talmy’s force-dynamic “tendency”
Consider the contrast between the following two causal scenarios:
(6) Onset and extended causation
a. Onset: The ball started rolling because of the wind.
b. Extended: The ball kept rolling because of the wind.
Talmy makes the basic distinction between onset and extended causation (1976, pp. 63–
66; 1985; 1988; 2000a, p. 417–419). In onset causation, Ant’s entrance or exit initiates
the effect whereas in extended causation, Ant’s force is exerting its influence onto Ago
during the entire conceptualization. (6a) is an example of onset causation. Thus the
appearance of wind initiates the rolling of the ball. (6b) is an example of extended
causation. Thus the wind is exerting its rolling influence onto the ball during the entire
duration of the conceptualized scene. As has already been discussed extensively in the
previous section, Talmy characterizes Ago and Ant as having intrinsic force tendencies.
A crucial point is that for Talmy the notion of “tendency” is a basic inherent feature
of Ago and Ant (Deane 1996; Talmy 2000a). This means that it is assumed that
tendency is a indispensable primitive of Ago and Ant—no matter if onset or extended
9
causation is involved. However, the main point to be made in the present section
is to point out that there is a conceptual problem when the notion of a “tendency” is
applied to onset causation (the previous section, where the notion of “tendency” featured
prominently, all contained examples of extended causation or at least examples where
an extended-causation interpretation was possible). To start investigating this issue,
let us first turn briefly to extended causation again where the notion of an “(intrinsic)
tendency” seems apt, using (6b) as an example. As mentioned, in (6b) the ball is
conceptualized as rolling during the entire scene, as we should expect in extended
causation. Consequently, the idea that the ball is somehow related to stationariness
can only be abstractly inferred. Thus, it makes perfect sense here to say that the ball
has a tendency toward stationariness—a tendency which it is not able to manifest at
any point because it is continuously suppressed by Ant’s impingement.
Let us now turn to onset causation. As Talmy (2000a, p. 417) remarks, in onset
causation the mental scene is basically segmented into two parts: the scene before the
Ant intervention and the scene after the Ant intervention. Thus in (6a) Ago is first
in a certain action or state before Ant’s impingement (that is, the ball is stationary)
and Ago is then in certain action or state after Ant’s impingement (that is, the ball
is rolling now). Note now, that before Ant’s impingement, Ago has not necessarily
abstract tendency toward stationariness: Ago simply is stationary—that is, in this
before-condition Ago is conceptualized to be in the state of stationariness rather than
having merely the tendency toward stationariness. It is only after Ant’s impingement,
where one can clearly characterize Ago to have the tendency toward stationariness.
Now Ago’s stationariness is not manifesting itself any longer in overt ways, but can
only be abstractly inferred.
We may therefore note that Talmy’s characterization of “tendency” does not seem
to be necessary in the before-impingement situation of onset causation. I would like to
10
suggest that Ago and Ant must necessarily be conceptualized to be in a given action or
state in this particular mental situation (in onset causation, before Ant’s impingement).
How can one incorporate these observations into the notational force-dynamic variant
of (5)? One solution is to remove “tendency” (T) from the description (cf. with (5)):
(7) C: Ago-x(–), Ant-xdiff(+) → E: Ago-xdiff
(7) for instance correctly predicts that (8) makes no sense:
(8) *The already rolling ball started rolling because of the wind.
This example demonstrates that the values of x and xdiff can in principle never be the
same, as captured by (7). Instead of standing for a “tendency,” x can now stand for
any value (any action or state) ascribed to Ago and xdiff can now stand for any value
(any action/state) ascribed to Ant, as long the values of Ago and Ant are not the
same. In relation to (6a) (onset causation), the cause (C) can now be identified as the
scenario before the impingement: Ago (here, the ball) is conceptualized to be in a given
action or state (x, here stationariness) and Ant (here, the wind) is conceptualized to
be in a different given action or state (xdiff, here motion). Ant imposes its stronger (+)
different value (xdiff, here motion) onto the weaker (–) Ago. This intervention has the
effect (E) that Ago takes on the imposed different value of Ant (motion). From this
it also follows logically that Ago can no longer manifest its initial value that it had
before the Ant intervention. Thus in (6a), in the postinterventional state, Ago’s initial
action/state value is suppressed—and now becomes a tendency. This observation of
a suppressed inherent value can also be logically derived from (7) without having to
explicitly incorporate the notion of “tendency” into the description. When compared
to Talmy’s solution of making “tendency” a primitive that holds throughout onset and
extended causation (2000a, p. 413–419), the present solution—(7)—has the advantage
that it avoids to speak of tendencies where the notion does not seem to be necessarily
11
involved (namely, in onset causation before Ant’s intervention). But how are we to
deal with the notion of “tendency” in extended causation? Should we simply add the
“T” (tendency) as a primitive again? I think that the systematic breaking up of force-
dynamic elements into a cause and into an effect (as has been done in Section (2.1))
will give us a more general solution, as we shall examine next.
2.3 Interpreting Talmy’s “onset/extended causation” distinction
in terms of cause and effect
Kant considered the contrast between onset and extended causation a long time ago:
If I consider a ball that lies on a stuffed pillow and makes a dent in it as
a cause, it is simultaneous with its effect. Yet I still distinguish the two by
means of the temporal relation of the dynamical connection. For if I lay
the ball on the pillow the dent follows its previously smooth shape; but if
(for whatever reason) the pillow has a dent, a leaden ball does not follow it
(1781/1998, p. 312).
Let us first consider the onset-causation example of Kant (“. . . if I lay the ball on the
pillow the dent follows its previously smooth shape”). This onset-causation scenario can
be captured with (7). The cause (C) involves conceptualizing Ago (the pillow) to be in
a given action/state x (smoothness) and conceptualizing Ant (the ball) as attempting
to impose a given action/state xdiff (denting) onto Ago. The ball intervenes on the
pillow’s smooth state by imposing its stronger (+) different value of denting onto the
weaker (–) pillow. This intervention has the effect (E) that Ago (the pillow) has to take
on the imposed different value (being dented) of Ant (the ball).
Let us now consider the extended -causation example that Kant examines (“If I con-
sider a ball that lies on a stuffed pillow and makes a dent in it as a cause, it is simulta-
12
neous with its effect”). In terms of cause and effect Kant aptly captures the very basic
distinction between onset and extended causation: in the former, cause and effect oc-
cur in sequence and in the latter, cause and effect occur simultaneously. Furthermore,
Kant also makes it clear that the conceptualization of extended causation requires the
assumption of an underlying onset-causation conceptualization (in extended causation
we “still distinguish the two [cause and effect] by means of the temporal relation of
the dynamical connection”). Thus, following Kant, onset causation is the unmarked
(default) case and extended causation is the marked case. Talmy also recognizes this
unmarked/marked distinction by calling onset causation the “prototypical form” (2000a,
p. 418). However, he does not develop this mentioning of “prototypical” status of onset
causation any further. The notational system developed in this article can capture the
unmarked/marked distinction between onset and extended causation in the following
way:
(9) Force dynamic elements in sequential and simultaneous occurrence of cause and
effect
a. Sequential cause and effect (onset causation): C: Ago-x(–), Ant-xdiff(+) → E:
Ago-xdiff
b. Simultaneous cause and effect (extended causation): [C: Ago-x(–),
Ant-xdiff(+) → E: Ago-xdiff]simul
(9b) is the same as (9a) except that in (9b) cause and effect are cognized simultaneously.
This is notated with [. . . ]simul which indicates that the entire cause-and-effect structure
is embedded within simultaneous cognition. For illustration we may put Kant’s two
examples into the notational description:
(10) Sequential and simultaneous occurrence of cause and effect with Kant’s
pillow/ball example
13
a. Sequential C→E: C: Agopillow-xsmooth(–), Antball-xdiff dented(+) → E:
Agoball-xdiff dented
b. Simultaneous C→E: [C: Agopillow-xsmooth(–), Antball-xdiff dented(+) → E:
Agoball-xdiff dented]simul
(10a) reads: the cause (C) involves cognizing an entity pillow (Ago) as being in an initial
smooth state (x) and as being weaker (–) than another cognized entity ball (Ant) which
is conceptualized as imposing the different action/state of denting onto the pillow.
Since Ant’s imposing force is conceptualized as being stronger (+) than Ago’s force,
the cognized effect (E) is that the pillow gets dented (E: Ago-xdiff).
(10b) reads: The cause (C) involves mentally construing a pillow (Ago) being in a
weaker (–) state of smoothness (x), a state which the stronger (+) ball (Ant’s inter-
vention (/) successfully changes by denting the pillow (xdiff). So far (10b) is identical
to (10a). But now this basic embedded sequential C→E relationship is additionally
cognized as occurring simultaneously. This process of simultanization allows the C→E
relationship to extend, so that Ant has the effect on Ago continuously over the entire
duration of the mental representation.
Coming back to Talmy’s notion of “tendency” (see Section 2.2), in the simultaneous
occurrence of cause and effect (extended causation) the situation for Ago is that its
inherent tendency is suppressed for the entire conceptualization. Thus, for instance
in Kant’s extended-causation example, the pillow’s tendency to have a smooth surface
is suppressed throughout the entire cognition of the scene. However, comparable to
the remarks in Section 2.2, this observation of a suppressed action or state can also
be logically derived from (9b) without having to explicitly incorporate the notion of
“tendency” into the notational system. Thus, as far as I can see, there is no need to
incorporate “tendency” as a conceptual primitive into force dynamics.
14
In this section, three basic concepts of Talmy’s force dynamics (1985; 1988; 2000a;
2011, pp. 633–634) have been revised: (i) the action/rest dichotomy has been replaced
with an x/xdiff dichotomy (Subsection 2.1); (ii) the notion that Ago and Ant have
“tendencies” has been replaced with the notion that they have “values” (actions or
states that are either expressed or suppressed (Subsection 2.2); and (iii) the onset/offset
distinction has been further refined in that it has been brought into an unmarked form
(sequential occurrence of cause and effect) and a marked form (simultaneous occurrence
of cause and effect) (Subsection 2.3). Let us then see how this somewhat revised account
of Talmy’s original force-dynamic account can handle successful causation (CAUSE,
Section 3), failed causation (DESPITE, Section 4), negative causation (PREVENT, 5),
and disengaged potential negative causation (ENABLE, Section 6).
3 Successful causation (s-causation)
One way of looking at constructions containing because—such as The ball started rolling
because of the wind—is to view them as successful causation (Jackendoff 1990, 131–133;
Jackendoff 1996, pp. 120–121). That is, a cause—such as the wind blowing onto a
ball—is successfully realizing the effect—such as making the ball roll. For convenience,
successful causation will be abbreviated to s-causation. Note now that (7) cannot only
capture the causal conceptual structure underlying because (of) (which linguistically
surfaces either as a subordinate conjunction or a subordinate preposition). (7) is also
readily used to analyze the conceptual structure underlying other linguistic forms that
express s-causation, as (11) suggests.
(11) Some linguistic variants of s-causation that can be described with (7)
a. Subordinate preposition: The ball started rolling because of the wind.
b. Subordinate conjunction: The ball started rolling because the wind blew on it.
15
c. Verb: The blowing of the wind caused the ball to start rolling.
d. Adverb: The wind blew. Therefore the ball started rolling.
All of the linguistic variations of expressing that a ball moves from the wind blowing
on it are readily analyzed with (7): They can all be described as involving a weaker
Ago (stationary ball) and a stronger Ant imposition (movement), which results in Ago
having to take on the different value of Ant. Thus I propose that (7) can stand for the
mental representation of s-causation in general. The concept of s-causation corresponds
to what sometimes is referred to as “CAUSE” (Sloman et al. 2009; Walsh & Sloman
2011; Wolff & Song 2003; Wolff 2007). I propose that (7) can capture s-causation
(which includes what is sometimes referred to as CAUSE), no matter how this concept
linguistically surfaces.
Thus far it has been demonstrated how (7) can be applied to the physical realm
(with the example of wind intervening on the condition of a ball). It has also been
demonstrated how (7) can deal with physical force-interactions that Talmy’s action/rest
dichotomy cannot handle (see (3)). In the remainder of this section it will be demon-
strated that (7) can also be applied to all sorts of phenomena that go beyond the
physical domain. Indeed, it seems to be the case that (7) is neutral to what kind of
conceptual content it can “handle.” As already pointed out by Talmy (1985; 1988;
2000a; 2011, p. 634) force-dynamic patterns are found in relation to all sorts of concep-
tual content (cf. also Wolff 2007). To give a sense of this content-neutrality, (12) gives
some examples of phenomena that go beyond the physical domain and where (7) seems
apt.
(12) s-causation in a variety of contexts that go beyond the purely physical domain
a. Human-physiological states: Mary went to bed because she was tired.
b. Social pressures: The tenor gave an encore because this was what the
16
audience expected of him.
c. Ideational: The argument is convincing because of its logical structure.
(12a) demonstrates that the x/xdiff dichotomy of (7) can be applied to totally differ-
ent causal scenarios—different to highly transparent physical motion examples (like
wind/ball examples)—where the Ago/Ant force interaction is perhaps more obscure:
for instance, a scenario where we conceptualize someone as going to bed because she
is tired. What is the causal force interaction that underlies a sentence such as Mary
went to bed because she was tired? (7) predicts that one should be able to identify an
Ago which must be in a condition (action or state) that can be conceived of as being
different to the condition of an Ant. When one is tired, it is only natural that one
wants to be somewhere where one can sleep, like being in bed. However, before going
to bed, one is not in such a place. Thus in (12a), Ago can be identified as Mary “being
out of bed” (that is, her “locational state” before she has actually walked to the bed and
lain down on it). Now the x/xdiff dichotomy suggests that we should find an Ant, in
this case a cognitive construct that has a different action/state value to “being out of
bed.” The solution seems to be that Ant is “Mary’s tiredness” which carries within it
the suggestion that it would be indicated to go to bed (linguistically implied in “went to
bed because she was so tired”). (7) can capture (12a) in the following way (the content
that appears in association with the s-causation elements has been added in underlined
subscript right after the element):
(13) C: AgoMary-xout of bed (–), AntMary’s tiredness-xdiff indication to go to bed (+) → E:
AgoMary-xdiff goes to bed
This reads: the cause (C) involves cognizing Mary’s current locational state (Ago-x:
Mary–out of bed) which is intervened by Mary’s tired state that carries within it the
suggestion that it would be indicated to go to bed (Ant-xdiff: Mary’s tiredness–indication
17
to go to bed). As in all s-causation, Ant is stronger (+), thus the effect (E) is that Mary
indeed goes to bed (E: Ago-xdiff: Mary–goes to bed).
(12b) demonstrates that the proposed x/xdiff dichotomy also easily works for socio-
dynamic metaphorical extensions of force dynamics (Talmy 2000a and Wolff 2007 have
both also demonstrated such force-dynamic extensions):
(14) C: Agotenor-xstop singing (–), Antexpectation of audience-xdiff give an encore (+) → E:
Agotenor-xdiff give an encore
This reads: the cause (C) involves cognizing the tenor’s current emotional state in
relation to interacting with the audience—the show is over and he does actually not
feel like giving an encore (Agotenor-xstop singing). But his Ago-state is intervened by him
cognizing that the audience has a different expectation: it expects him to give an encore
(Antexpectation of audience-xdiffgive an encore). Ant is stronger (+), thus the effect is that the
tenor gives the audience an encore (E: Agotenor-xdiff give an encore).
(12c) demonstrates that the proposed x/xdiff force-dynamic dichotomy can also be
extended to abstract realms where Ago and Ant are cognized as ideational entities.
Additionally, the most likely reading of (12c) is the one where cause (C) and effect (E)
occur simultaneously (cf. Section 2.3):
(15) [C: Agoargument-xdoubtful (–), Antlogical structure-xdiff convincing(+) → E:
Agoargument-xdiff convincing]simul
The underlying sequential cognition of cause and effect ([C→E]) reads: the cause (C)
involves cognizing the argument as being initially cognized as doubtful (Agoargument-
xdoubtful). (If the argument were not construed as possibly being initially doubtful,
there would be no reason why one would have to give a reason for its convincibilty.)
This possible initial doubtfulness is intervened by cognizing that the argument does
18
have a logical structure, which carries within it the force of convincibility (“is convinc-
ing because of its logical structure”; Antlogical structure-xdiffconvincing ). As in all s-causation,
Ant is stronger (+), thus the effect is that the argument is cognized as convincing (E:
Agoargument-xdiffconvincing). Additionally, the cognitive process of cause-and effect simul-
tanization ([C→E]simul) allows the C→E relationship to extend, so that Ant (the logical
structure) has the effect (of being convincing) on Ago (the argument) continuously over
the entire duration of the mental representation. A sequential occurrence of cause and
effect in relation to the basic ideational content of (12c) would of course also be pos-
sible (for instance underlying a sentence like The argument was eventually convincing
because of its logical structure).
The examination undertaken in this section has provided evidence that we mentally
dichotomize s-causation (CAUSE) into a cause—C: Ago-x(–), Ant-xdiff(+)—and an
effect—E: Ago-xdiff.
4 Failed causation (f-causation)
In the previous section mental causality has been examined in relation to successful
causation (s-causation). There is also the opposite: the instance of what one might
view as “failed causation” (Jackendoff 1990, pp. 131–133; Jackendoff 1996, pp. 120–
121). In such a case the cause—such as the wind blowing onto a ball— fails to realize
the potential effect—fails for instance to make a ball roll. Failed causation will be called
f-causation. What linguistically surfaces when expressing f-causation is for instance the
preposition despite (Talmy 1976, 1985, 1988; 2000a; 2000c; Wolff 2007; Wolff & Song
2003).
(16) Contrasting successful and failed causation
a. Successful: The ball started rolling because of a sudden gust of wind blowing
19
on it.
b. Failed: The ball remained still despite a sudden gust of wind blowing on it.
The newly developed x/xdiff dichotomy allows one not only to systematically describe
s-causation (Section 3), it also provides a means to describe f-causation, as for instance
underlying the use of despite. This is so because the conceptual structure underlying
because and the conceptual structure underlying despite contain the same elements,
with the only difference that in the former Ant is stronger than Ago and in the latter
Ago is stronger than Ant (Talmy 2000a, pp. 415–417). Using the notational variant
that has been introduced in the previous sections, f-causation can be captured in the
following way (cf. with (7) in Section 2.2):
(17) Force-dynamic elements of failed causation (f-causation)
C: Ago-x(+), Ant-xdiff(–) → E: Ago-x
Thus in (16b), the (failed) cause (C) is: a stronger (+) Ago (the ball) with some
given action/state (x: stationariness) resisting the imposition of a different action state
(xdiff: motion) of a weaker (–) Ago. Given that Ant is weaker, the (failed) effect (E) is
consequently in accordance with Ago’s initial value x—the ball’s (Ago’s) inertial force
is stronger than the wind’s (Ant’s) pushing force and is therefore not moved by it.
(17) can also be used to analyze the conceptual structure underlying other linguistic
forms that express f-causation, as (18) suggests.
(18) Some linguistic variants of f-causation that can be described with (17)
a. Subordinate preposition: The ball remained still despite a sudden gust of
wind blowing on it.
b. Subordinate conjunction: The ball remained still although a sudden gust of
wind blew on it.
20
c. Adverb: A sudden gust of wind blew on the ball. Nevertheless the ball would
not move.
All of the linguistic variations of expressing that a ball does not move despite a sudden
gust of wind blowing on it, are readily analyzed with (17): They can all be described
as involving a stronger Ago (ball) intervened by a weaker Ant (gust of wind), which
results in Ago not having to take on the different value of Ant. Thus I propose that (17)
can stand for the mental representation of f-causation in general. f-causation is some-
times also referred to as “DESPITE” (Wolff & Song 2003; Wolff 2007). Thus I propose
that (17) can also stand for DESPITE. Further evidence for this proposition is given
in (19), which demonstrates that (17) can adequately describe f-causation in a large
variety of conceptual contexts that go beyond the physical domain (these f-causation
examples are systematically contrasted with the s-causation examples of (12) in Sec-
tion 3).
(19) f-causation in a variety of conceptual contexts
a. Human-physiological states: Mary had to stay up although she was tired.
b. Social pressures: The tenor decided not to give an encore although this was
what the audience expected of him.
c. Ideational: The argument is doubtful despite its logical structure.
(19a) can be notated as (cf. (13)):
(20) C: AgoMary-xout of bed (+), AntMary’s tiredness-xdiff indication to go to bed (–) → E:
AgoMary-xout of bed
This reads: the failed cause (C) involves cognizing Mary’s current locational state (Ago-
x: Mary–out of bed) which is intervened by Mary’s tired state that carries within it
21
the suggestion that it would be indicated to go to bed (Ant-xdiff: Mary’s tiredness–
indication to go to bed). But as in all f-causation, not Ant but Ago is stronger (+).
Thus the effect of Mary going to bed fails—she has to (for some unspecified reason)
stay out of it (E: Ago-x: Mary–out of bed).
Turning to (19b), this can be notated as (cf. (14)):
(21) C: Agotenor-xstop singing (+), Antexpectation of audience-xdiff give an encore (–) → E:
Agotenor-xstop singing
This reads: the failed cause (C) involves cognizing the tenor’s current emotional state
in relation to interacting with the audience—the show is over and he does actually
not feel like singing an encore (Agotenor-xstop singing). His Ago-state is intervened by him
cognizing that the audience has a different expectation: it expects him to give an encore
(Antexpectation of audience-xdiffgive an encore ). However, his Ago-state is stronger (+). Thus
the effect is that the tenor decides not to sing an encore (E: Agotenor-xdiffsing an encore),
despite the audience expecting him to do so.
Finally, turning to (19c), this can be notated as (cf. (15)):
(22) [C: Agoargument-xdoubtful (+) / Antlogical structure-xdiff convincing(–) → E:
Agoargument-xdoubtful]simul
In (19c) the underlying sequential cognition of cause and effect ([C→E]) reads: the failed
cause (C) involves cognizing the argument as being doubtful (Agoargument-xdoubtful). This
is intervened by cognizing that the argument does have a logical structure. While this
logical structure is conceptualized to carry within it the possible force of convincibility,
this force nevertheless fails to have an effect, since the initial impression that the ar-
gument is doubtful remains stronger. (An argument might for instance have a logical
structure, but remains doubtful for other reasons—e.g., ethical reasons.) Additionally,
the cognitive process of cause-and-effect simultanization ([C→E]simul) allows the failed
22
C→E relationship to extend continuously, so that Ant (the logical structure) has no
(convincing) effect on Ago (the argument) over the entire duration of the mental rep-
resentation. A more sequential occurrence of failed cause and effect in relation to the
basic ideational content of (19c) also seems possible (for instance underlying a sentence
like The argument started to become doubtful despite its logical structure).
The examples of (19) suggest that (17)—analog to (7) in Section 2.2—is neutral
as to what kind of conceptual content can be conceptualized within it. As long as
we conceptualize a “failed cause-and-effect relationship” within this mental scaffolding,
then (17) is “willing” to let content from any given conceptual realm splay across it.
The examination undertaken in this section has provided evidence that we mentally
dichotomize f-causation (DESPITE) into a “failed cause”—C: Ago-x(+), Ant-xdiff(–)—
and a “failed effect”—E: Ago-x.
5 Negative causation (n-causation)
There is also one particular kind of cause-and-effect relationship that has received much
separate attention in the literature: the concept of “PREVENT” (Sloman et al. 2009;
Talmy 1985, 1988; 2000a; Walsh & Sloman 2011; Wolff 2007; Wolff & Song 2003).
PREVENT is often characterized as “negative causation” (e.g., Barbey & Wolff 2007;
Walsh & Sloman 2011; it will thus be abbreviated to n-causation in this article). When
contrasting n-causation with s-causation (Section 3), then s-causation (“successful cau-
sation”) can also be viewed as “positive causation”:
(23) Contrasting positive and negative causation
a. Positive: The ball started rolling because of a sudden gust of wind blowing on
it.
b. Negative: The ball (on the incline) could not start rolling because of a stone
23
lying in the way.
(23a) is “positive” in the sense that conceptualizing this kind of causality does not
involve conceptualizing negation (presumably not even implicitly): The effect, the ball
rolling, is conceptualized in positive terms—that is, no negation is added to the concept
of rolling. In contrast,(23b) is “negative” in the sense that conceptualizing this kind
of causality does involve conceptualizing negation: the effect—the ball not rolling—is
conceptualized in negative terms—that is, negation (here in the form of not) is added
to the concept of rolling.
Thus, when contrasting it with n-causation, it seems natural to characterize regu-
lar causation (Section 3; for instance the conceptual structure underlying because) as
positive causation. But when regular causation is contrasted with f-causation (Sec-
tion 4; for instance the conceptual structure underlying despite) it seems more natural
to characterize it as successful causation (as has been done in Section 3). Given that
a cognitive-linguistic analysis acknowledges, for instance, “partial overlaps” or “interac-
tions that lead to mutual modifications” (Talmy 2011, p. 624; cf. Section 1) this presents
no problem. Rather than having to fix our mind on whether we shall best characterize
regular causation (because, to cause and so on) as successful or positive, we can state
that this might differ depending on what regular causation is contrasted with. Thus,
while I will keep to refer to regular causation as successful causation (s-causation), one
should bear in mind that, depending on context, it might just as well be called “p-
causation” (positive causation). Allowing for this degree of freedom in the descriptive
framework is perhaps a good example how the current cognitive-linguistic approach to
force dynamics differs from a more typical algebraic account like Jackendoff’s concep-
tual semantics. He adopts a fixed notion of “successful causation” with the use of “the
notation CS+ to encode application of force with a successful outcome” (1990, p. 132).
But, as we have seen, linguistic evidence suggests that we might—as this article does—
24
need a more dynamic description of regular causation, one that can encode either the
notion of success or of positivity.
In (23b), n-causation is expressed with “not X because Y” (not rolling because of a
stone) (cf. Walsh & Sloman, 2011). The conceptual structure of “not X because Y” can
also be captured with the notational variant of Talmy’s force-dynamic diagramming
system (Talmy 2000a, p. 414) that has been developed thus far in this article. Recall
that so far Ago has been characterized as having any (overt or suppressed) action/state
value x and Ant as having an action/state value that differs from Ago’s value (xdiff)
(Sections 3–4). One may note that when contrasting an action/state with the negation
of that action/state, that then this negated action/state can also be viewed as a kind of
difference: doing something and not doing something are, after all, two different kinds
of scenarios. In n-causation the value of Ant differs from the value of Ago in always the
same way: Ant always “wants” a result that negates the initial value of Ago. For these
reasons I will replace difference (diff ) with negation (neg) in capturing n-causation
notationally (cf. with (7) in Section 2.2):
(24) Force-dynamic elements of negative causation (n-causation)
C: Ago-x(–), Ant-xneg(+) → E: Ago-xneg
Considering (24) a theorist who would want to stick to a more typical algebraic nota-
tional system might ask: Can one simply replace diff with neg? But in a cognitive-
linguistic analysis in the sense of Talmy (2011, p. 624; cf. Section 1), it would perhaps
even be a surprise if a rigid notion of difference would characterize all tendencies of
an Ant. Instead, we find, as one expects in a cognitive-linguistic analysis, that a very
broad, somewhat vague notion of difference applies, one that includes negation within
its scope. Such a notion of negation, viewing it as a specific case of “differentness” also
differs from the concept of negation in propositional logic where the notion of negation
is not explicitly linked to the notion of “differentness”.
25
In (24) the “negative cause (negative C)” (an action/state which will result in
negating some action/state of Ago) is represented as a weaker (–) Ago with a given
action/state x and a stronger (+) imposing Ant with the value to negate this ac-
tion/state x of Ago (xneg). The “negative effect (negative E)” on Ago is represented as
“Ago-xneg”: Ago has to take on Ant’s negation and is thus “prevented” from acting out
its its initial state or action.
Furthermore, in (23b), the negative effect is present during the entire conceptual-
ization, thus C and E occur simultaneously ([C→ E]simul; cf. Section 2.3):
(25) Simultaneous occurrence of negative cause and effect
[C: Ago-x(–), Ant-xneg(+) → E: Ago-xneg]simul
(23b) can also be described as an “extended negative onset” cause-and-effect relation-
ship: The negative causation (preventing the ball from moving) extends continuously—
the stone exerts its blocking force onto the ball on the incline continuously—and contin-
uously prevents the ball from starting to roll. Hence negated onset causation is extended
repeatedly over the entire duration of the mental representation. Thus we may note
that the notions of extendedness and onsetness can feature within a single mental repre-
sentation of causation. This is an observation which Talmy’s onset/extended causation
dichotomy (1976, pp. 63–66; 1985; 1988; 2000a, p. 417–419) does not seem to predict,
since Talmy characterizes mental representation of causation as occurring in either on-
set or extended form. This conflation of onsetness/extendedness in a single mental
representation is less directly relevant to the present article, since this onset/extended
distinction has been captured as the difference between sequential or simultaneous oc-
currence of cause and effect—which does not result in a comparable conflation.
Thus far in this section, (24) has been examined in relation to the linguistic con-
struction “not . . . because” (“not X because Y”). However, the same description can
26
readily be applied to refer to the verb form that expresses roughly the same concept,
to the verb prevent (from):
(26) Two linguistic variants of n-causation
a. Subordinate preposition and negation adverb: The ball (on the incline) could
not start rolling because of a stone lying in the way.
b. Verb: The stone prevented the ball (on the incline) from starting to roll.
The conceptual structure of both linguistic variants are readily described with (24)
or (25): They can both be described as involving a weaker Ago state (the ball’s sta-
tionariness) that is intervened on by a stronger negation Ant (stone), which results in
Ago having to take on the negative value of Ant.
Thus I propose that (24)—or if the negative cause and effect occur simultane-
ously: (25)—can stand for the mental representation of the concept of “prevention”
in general. The concept for the mental representation of prevention is often referred
to as “PREVENT” (cf. Sloman et al. 2009; Walsh & Sloman 2011; Wolff & Song 2003;
Wolff 2007). Thus I propose that (24) can stand in general for n-causation or what is
often characterized as PREVENT.
Further evidence for the proposal that (24) can stand quite generally for the concept
of n-causation is given in (27) which demonstrates that (24) can adequately describe
n-causation (or PREVENT) in a large variety of conceptual contexts. This is demon-
strated with both examined linguistic variants of PREVENT, with ”not X because Y”
and “prevent (from)” (these n-causation examples are systematically contrasted with the
s-causation examples of (12) and with the f-causation examples of (19), in Sections 3
and 4, respectively):
(27) n-causation in a variety of conceptual contexts that go beyond the purely physical
domain
27
a. Human-physiological states: Mary could not fall asleep because she had had
two double espressos. / Having had two double espressos prevented Mary from
falling asleep.
b. Social pressures: The tenor did not give an encore because the audience was
booing. / The booing of the audience prevented the tenor from giving an encore.
c. Ideational: The argument was not convincing because it was lacking a logical
structure. / Lack of a logical structure prevented the argument from being
convincing.
(27a) can be notated as (cf. (13)):
(28) C: AgoMary-xintends to fall asleep (–), Antdouble espressos-xneg not fall asleep (+) → E:
AgoMary-xneg not fall asleep
This reads: the cause (C) involves cognizing Mary’s intentional state (Ago-x: Mary–
intends to fall asleep) which is intervened by Mary’s having had two double espressos.
These double espressos are experienced as carrying within them the power of negating
the possibility of falling asleep at the moment (Ant-xdiff: double espressos–not fall
asleep). As in all n-causation, the negating Ant is stronger (+), thus the effect is that
Mary is indeed unable to fall asleep (E: Ago-xdiff: Mary–not fall asleep).
(27b) can be notated as (cf. (14)):
(29) C: Agotenor-xfeels like giving an encore (–), Antbooing of audience-xneg not give an encore (+) →
E: Agotenor-xneg not give an encore
This reads: the cause (C) involves cognizing the tenor’s current emotional state in re-
lation to interacting with the audience—the official performance is over and he at first
actually feels like giving the audience an encore (Agotenor-xfeels like giving an encore). But
this initial Ago-state is intervened by noticing the booing of the audience, which carries
28
within it the suggestion that he should refrain from giving an encore (Antbooing of audience-
xneg not give an encore ). Ant is stronger (+), thus the effect is that the tenor (Ago) is pre-
vented from acting out his initial desire to sing an encore (E: Agotenor-xneg not give an encore).
Finally, turning to (19c), this can be notated as (cf. (15)):
(30) C: Agoargument-xconvincing (–), Antlack of logical structure-xneg not convincing(+) → E:
Agoargument-xneg not convincing
This reads: the cause (C) involves cognizing the possibility that the argument could
be convincing (Agoargument-xconvincing). But noticing that the argument is lacking a
logical structure (Antlack of logical structure) eventually prevents the argument from being
convincing (E: Agoargument-xneg not convincing).
The examination undertaken in this section has provided evidence that we mentally
dichotomize n-causation (PREVENT) into a “negative cause”—Ago-x(–), Ant-xneg(+)—
that has a “negative effect”—Ago-xneg.
6 Disengaged potential negative causation (dpn-causation)
In Section 5 the mental representation of n-causation has been characterized. In this
section, analysis is carried out which suggests that the concept of “ENABLE” (Sloman
et al. 2009; Talmy 1976, 1985, 1988; 2000a; 2000c; Wolff 2003, 2007; Wolff & Song
2003) can be characterized as “disengaging the potential effect of negative causation”—
that is, as disengaging a potential n-causation (PREVENT) scenario (cf. Talmy, 2000a,
p. 421 and Section 7.2). When contrasting “disengagement of potential negative causa-
tion” (which will be abbreviated to dpn-causation) with what has been characterized as
n-causation or PREVENT (Section 5), n-causation can also be viewed as “realized neg-
ative causation.” Contrasting the following sentences brings out the proposed contrast
between n-causation (PREVENT) and dpn-causation (ENABLE):
29
(31) Contrasting realized negative causation and disengaged potential negative
causation
a. Realized negative causation (n-causation): The ball (on the incline) could not
start rolling because of a stone lying in the way.
b. Disengagement of potential negative causation (dpn-causation): The ball that
potentially could not have started rolling because of a stone lying in the way
could start because someone removed the stone.
c. Disengagement of potential negative causation rephrased in terms of
ENABLE: Removing the stone enabled the ball to start rolling.
(31a) represents “realized negative causation” in the sense that conceptualizing this kind
of causality actually results in the negative effect: the ball is in effect conceptualized as
not rolling. In contrast, in (31b–c) “negative causation” is only a potential. The actual
result is that a potential prevention is disengaged : removal of the stone lets the ball
roll on.
The conceptual structure underlying dpn-causation can also be captured with the
notational characterization developed thus far in this article—as long as we allow to add
“potentiality” (pot) and “disengagement” (disen) as cognitive elements (cf. with. (7), (17),
and (24), in Sections 2.2, 4, and 5, respectively):
(32) Force-dynamic elements of disengaged potential negative causation
(dpn-causation)
a. dpn-causation (ENABLE)
C: [[C: Ago-x(–), Ant-xneg(+) → E: Ago-xneg ]pot ]disen → E: Ago-x
b. dpn-causation (ENABLE) formulated with the n-causation (PREVENT) part
not written out:
30
C: [PREVENTpot ]disen → E: Ago-x
When dpn-causation is—as in (32a)—fully written out, then a sense is provided of the
proposed conceptual complexity that is involved when one conceptualizes ENABLE.
However, since a large part of this characterization represents the characterization
of n-causation (PREVENT, (24)), one can also notate dpn-causation simply as done
in (32b), which presupposes PREVENT (see Section 5). Thus (32) proposes that dpn-
causation—based on the linguistic observations in (31)—in principle involves the con-
ceptualization of an embedded potential PREVENT scenario. In (31b–c), for instance,
the ball could potentially not roll because of a stone lying in the way. In ENABLE,
however, such potential negative causation is always made ineffective by disengaging
(disen) it. Thus in (31b–c) the disengagement of potential negative causation mani-
fests itself as someone removing the stone. Since the whole potential negative effect
of the negative causation is annihilated (disengaged), Ago can act out its initial ac-
tion/state x—or the disengagement gives Ago at least the potentiality to now act out
its initial value (see further below in section). In (31b–c), for instance, the effect of the
disengagement is that the ball can act out its initial Ago-action of rolling.
Evidence for the proposal that (32) can in principle adequately describe the con-
ceptual structure of dpn-causation (ENABLE) is given in (33), which demonstrates
that (32) can adequately describe dpn-causation or ENABLE in a large variety of con-
ceptual contexts (these dpn-causation examples are systematically contrasted with the
s-causation examples of (12), the f-causation examples of (19), and the n-causation
examples of (27), in Sections 3, 4, and 5, respectively):
(33) dpn-causation in a variety of conceptual contexts that go beyond the purely
physical domain
a. Human-physiological states: Mary who potentially could have been unable to
fall asleep because she had had two double espressos, could fall asleep because a
31
few hours had passed since she had had them. / Allowing enough time to elapse
between having two double espressos and going to bed enabled Mary to fall
asleep.
b. Social pressures: The tenor who may have potentially felt unable to give an
encore because of some booing in the audience could sing an encore because he
realized that most people were applauding. / Realizing that many more people
were applauding than booing enabled the tenor to be in the right mood to sing
an encore.
c. Ideational: The argument that was potentially unable to proceed with
clearness because it seemed at first to lack a logical structure could proceed with
clearness because the (initially intransparent) logical structure was made
transparent. / Making the logical structure of the argument transparent enabled
it to proceed with clearness.
All these instances involve (i) an Ago with a given initial action/state; (ii) an Ant with
a potential to negate (un-) the initial x-value of Ago; (iii) a potential intervention on
a weaker Ago by a stronger Ant; (iv) the disengagement of such potential negative
causation; and (v) an Ago that can act out its initial action/state value due to the
disengagement of the potential negative causation. Thus all these instances can be
aptly characterized with (32), the proposed conceptual structure of dpn-causation or
ENABLE. What still must be dealt with, however, is a refinement of (v): while it is
true that Ago is now in a position to act out its initial action/state value due to the
disengagement of the potential negative causation, it is not always the case that Ago
actually manifests its initial Ago action or state. This becomes particularly clear when
Ago is an agent: a person who is enabled to do something may or may not engage in
this action. As Talmy puts it, the actual occurrence of an effect is “not entailed in the
usual reading of the enable verbs” (1976, p. 508). The following examples bring out
32
this contrast:
(34) dpn-causation with and without the occurrence of an effect
a. dpn-causation with the occurrence of an effect: Realizing that many more
people were applauding than booing enabled the tenor to be in the right mood
to sing an encore [and so he went ahead and gave an encore].
b. dpn-causation without the occurrence of an effect: Realizing that many more
people were applauding than booing enabled the tenor to be in the right mood
to sing an encore [but remembering that he should not overuse his voice, he
nevertheless decided against giving an encore].
In other words, enabling empowers the enabled one to bring about the effect, but the
enabled one might still decide to not bring about the effect. What is needed notationally
for this conceptual refinement then is that the effect either takes place or remains only
a potential. This can be characterized with the notational system developed in this
article in the following way:
(35) dpn-causation with and without the occurrence of an effect
a. dpn-causation with the occurrence of an effect:
C: [PREVENTpot ]disen → E: Ago-x
b. dpn-causation without the occurrence of an effect:
C: [PREVENTpot ]disen → [E: Ago-x]pot
For (35a), the effect is not marked as a mere potential and thus takes place (E: Ago-
x). In (34a) for instance, the potential negative causation (the booing of some in
the audience that could potentially have prevented the tenor from feeling able to give
an encore) is disengaged (by others in the audience who are applauding). The tenor
33
decides to use this empowerment to have an effect and thus he gives an encore (E:
Agotenor-xgives an encore).
For (35b), the effect is marked as only being a potential ([E: Ago-x]pot). Accordingly,
the effect remains only an unrealized possibility. In (34b) for instance, the potential
negative causation (the booing of some in the audience that could potentially have
prevented the tenor from feeling able to give an encore) is—as in (34a)—still disengaged
(by others in the audience who are applauding). However, the tenor decides not to use
this empowerment to bring about the effect (of giving an encore). Thus the effect
remains only a potential ([E: Agotenor-xgives an encore ]pot).
The examination undertaken in this section has provided evidence that we mentally
dichotomize dpn-causation (ENABLE) into a “disengaged potential negative cause”—
[PREVENTpot ]disen—that either has the overall effect of letting Ago act out its initial
value—E: Ago-x—or that at least lets this effect exist as a potential.
7 Discussion
As has been demonstrated in the previous sections (Sections 2–6), the cause-and-effect
structure of four major forms of mental causality—s-causation (CAUSE), f-causation
(DESPITE), n-causation (PREVENT), and dpn-causation (ENABLE-)—can be argued
to involve the following cause (C) and effect (E) descriptions:
(36) Elementary descriptions for four major forms of cause and effect
• s-causation: C: Ago-x(–), Ant-xdiff(+) → E: Ago-xdiff
• f-causation: C: Ago-x(+), Ant-xdiff(–) → E: Ago-x
• n-causation: C: Ago-x(–), Ant-xneg(+) → E: Ago-xneg
• dpn-causation with E occurrence: C: [PREVENTpot ]disen → E: Ago-x
34
• dpn-causaton without E occurrence:C: [PREVENTpot ]disen →[ E: Ago-x]pot
Furthermore, C and E can also occur simultaneously (Sections 2.3, 4, and 5), which for
instance in relation to s-causation (Section 2.3) is represented in the following way:
(37) [C: Ago-x(–), Ant-xdiff(+) → E: Ago-xdiff]simul
If one breaks down these descriptions of mental cause and effect into their conceptual
primitives, then one derives at the following inventory of the elements that make up
cause and effect:
(38) The elements of cause and effect (of s-causation, f-causation, n-causation,
dpn-causation)
• agonist (Ago)
• antagonist (Ant)
• any state/action (x)
• difference (diff)
• stronger than (+)
• weaker than (–)
• negation (neg)
• potentiality (pot)
• disengagement (disen)
• sequentiality (→)
• simultaneity ([→]simul)
These eleven elements arrange themselves in the specific ways that have been described
(Sections 2–6) to provide the cause-and-effect structure of s-causation, f-causation, n-
causation, and dpn-causation. Some of these elements cluster together to form a cause
35
(C)—a successful cause, a failed cause, a negative cause, or a disengaged potential
negative cause. And a few elements cluster together to form the conceptually much
more simply structured effect (E)—a successful effect, a failed effect, a negative effect,
or a disengaged potential negative effect.
Having this mental inventory of the elements of major forms of causation at hand
now, let us in the present discussion section examine some further aspects of mental ele-
mentary causality. First the contrast between digital and analog cognition (Section 7.1)
and the concept of concordance (Section 7.2) are examined in a causal context. Then
the question how the elements of cause and effect feature in epistemics (Section 7.3) and
in counterfactual and probabilistic reasoning (Section 7.4) is addressed. Finally, it is
discussed how these force-dynamic elements might manifest in spatiotemporal structure
(Section 7.5).
7.1 Digital and analog representation in mental causality
Deane has stated that Talmy’s force dynamic account is “implicitly analog” (1996, p. 57).
In this context he envisions what a force-dynamic theory of mental causality must also
be able to account for in analog terms:
From a cognitive perspective, Talmy’s theory is a striking example of a
psychologically plausible theory of causation. Its key elements are such
concepts as the (amount of) force exerted by an entity, the balance between
two such forces, and the force vector which results from their interaction.
Such concepts have an obvious base in ordinary motor activities: the brain
must be able to calculate the force vector produced by muscular exertion,
and calculate the probable outcome when that force is exerted against an
object in the outside world . . . the amount of force exerted may vary con-
tinuously; the forces may be directly (or partially) opposed, or they may
36
coincide partially or in full. The state resulting from a force-dynamic inter-
action may vary both in the direction and magnitude of the resultant forces
(1996, pp. 56–57).
While Deane makes the valuable point that analog aspects are also important for force-
dynamic theory, his assumption that Talmy’s force dynamic account (1985; 1988) can
capture such mental analog calculations does not seem correct. Take Talmy’s system
of force-dynamic diagramming (1985; 1988; 2000a; as presented in Section 2.1). I know
of no way how these notational elements—Ago, Ant, , >, (+), (–), and so on—
could for instance represent how “the amount of force exerted may vary continuously.”
Accordingly, it seems that Jackendoff is right when stating that “Talmy’s [force-dynamic]
representations are as symbolic as mine” (1996, p. 122; cf. also Jackendoff 1990, pp. 130–
150). The Talmyan variant of force dynamics developed in this article—which revised
three basic concepts of Talmy’s force dynamics (Sections 2.1–2.3)—is also as symbolic
(or digital) as the accounts of Talmy or Jackendoff (Stocker, 2013). As also pointed
out by Stocker (2013), future research could determine how digital aspects of force-
dynamic causality (Jackendoff, 1990, 1996; Stocker, 2013; Talmy, 2000a) interact with
analog (vectorial) aspects of force-dynamic causality (Talmy, 1976, pp. 62–63; Barbey
& Wolff, 2007; Wolff, 2003, 2007; Wolff & Song, 2003; Wolff & Zettergren, 2002).
7.2 Basic opposition vs. basic opposition/concordance
Wolff and Song write:
. . . in Talmy s (1988) account, ‘‘underlying all more complex force-dynamic
patterns is the steady-state opposition of two forces. . .’’ (p. 53). Thus, in
Talmy’s theory the vast majority of force dynamic interactions are said to
involve opposition. However, in our account, we follow Jackendoff’s (1990)
37
observation that many such complex interactions appear to involve concor-
dance rather than opposition between the affector and the patient (2003,
p. 283).
Thus, one may divide the theoretical approaches of force dynamics into two broad
categories: in the former, all complex force-dynamic interactions basically involve op-
position (Talmy 1985; 1988; 2000a; and the Talmyan variant outlined in this article); in
the latter, complex force-dynamic interactions can basically either involve opposition
or concordance (Jackendoff 1990, pp. 133–134; Wolff & Song 2003; Wolff 2007). Jack-
endoff and Wolff base their analyses of concordance on ENABLE or ENABLE-related
analyses. Indeed, they treat ENABLE (or what is called dpn-causation in this article)
as a basic case of concordance. However, as has been given evidence for in Section 6, an
underlying opposition (difference) is clearly there as a potential in ENABLE. As Talmy
remarks in general in relation to disengaged force-dynamic interaction:
The notions of Agonist and Antagonist, it can be argued, intrinsically in-
volve the engagement of two bodies in an opposition of force, and reference
to an Agonist and Antagonist not so engaged necessarily depends on their
potential for such engagement (2000a, p. 421).
One can note that ENABLE generally involves the notion that a certain result cannot
be reached without aid, without the aid of the enabling entity—an observation that is
also in accordance with Wolff (2007, p. 93). Thus, for instance in (31b–c) (Section 6) the
stone’s rolling can only result from some external aid. In conceptualizing dpn-causation
or ENABLE, this “aid” is realized through the examined disengagement of the potential
negative causation. It is this disengagement that is in concordance with Ago’s initial
value. However, there is still the potential underlying Ago-Ant opposition: What is not
considered in the analyses of ENABLE of Jackendoff and Wolff is that the very idea
38
that some result cannot be reached without aid, logically requires that there must at
least be potential opposition to the desired result (as proposed in Section 6)—because
if there was no potential opposition whatsoever, then there could be no reason why aid
should be required to begin with. Thus the challenge for the force-dynamic theories
of Jackendoff and Wolff is to explain how ENABLE can make sense without assuming
some underlying potential opposition.
7.3 Force-dynamic elementary causation and epistemics
Consider:
(39) Mary must have gone to bed early, because she was so tired.
(39) is an inferred version of (12a)—of Mary went to bed because she was tired (Sec-
tion 3). Perhaps the most basic distinction in epistemics is whether a speaker knows
or infers a stated proposition (Talmy in Ibarretxe Antuñano 2006, pp. 260–261). Thus
a speaker uttering (12a) treats this proposition as “factual s-causation”. In contrast, a
speaker uttering (39) (let’s call this speaker John) infers that this causal proposition is
very likely to have occurred—John mentally engages in “inferred s-causation”.
A cognizer making an inference must also have some information in mind that she or
he treats (implicitly or explicitly) as factually unknown. Only when there is something
treated as factually unknown can there be a contrast with information that is inferred.
The piece of information that John implicitly assumes as factually unknown in (39) is
that he does not know if Mary went to bed early. If he knew it as a fact, then there
would be no need for an inference. It is the very act of inferring that implies that the
matter that is inferred upon is not known to a speaker. In order for an inference to occur
in such a state of unknowing, it seems plausible to assume that there must be some
cognitive trigger that prompts the making of an inference. In (39) this trigger for John is
39
Mary’s tiredness (“because she was so tired”). So the basic pieces of mental information
that John contrasts are: his factual unknowing if Mary went to bed early versus his
inferring that she must have gone to bed early, the latter being prompted by knowing
of Mary’s given tiredness. One may also notice that the cognitive trigger itself—Mary
being tired—is treated as fact by John. s-causation can provide an apt explanatory
framework of how inference allows one to override factual unknowing (cf. with (7) in
Section 2.2):
(40) C: AgoMary-xunknown if went to bed early (factual) (–),
AntMary’s tiredness-xdiff went to bed early (inferred)(+) → E:
AgoMary-xdiff went to bed early (inferred)
This reads: the cause (C) first involves John cognizing that he in fact does not know if
Mary went to bed early. Since this factual value will not be the main resultant value, it
can be viewed as weaker than the inferential value that will be the resultant value. Ac-
cordingly the fact of unknowing can be treated as a weaker (–) Ago-force of s-causation
(Ago-x: Mary–unknown if went to bed early (factual)). This cognition of factual un-
knowing is then intervened in John’s mind by knowing of Mary’s tiredness, which serves
as a trigger to build up an inferential force suggesting to him that she must have gone
to bed early. Thus the act of inferring can be treated as the stronger (+) Ant-force
(Ant-xdiff: Mary’s tiredness–went to bed early (inferred)). Since Ant is stronger (+),
the effect (E) is that John does not simply state that he does not know for certain if
Mary went to bed early, which also would be a possible statement after all. Rather
John indeed infers that Mary went to bed early (E: Ago-xdiff: Mary–went to bed early
(inferred)).
Future research could determine if all inferred causal statements can be analyzed as
involving a weaker Ago-force of not knowing about a situation that is overruled by a
stronger Ant-trigger that allows one to infer something about a situation that in actual
40
fact is unknown.
7.4 Force-dynamic elementary causation and counterfactual and
probabilistic theories of causation
The account of force-dynamic cause and effect, as developed in this article, is fully
compatible with both counterfactual and probabilistic accounts of causality (such as
the ones found in Pearl 2000; Sloman 2005; Sloman et al. 2009; Spirtes et al. 2000). Let
us turn to counterfactuals first.
A counterfactual criterion of causation states that an event can only be viewed as
the cause to bring about an effect on another event, if it is the case that if cause had
not occurred, then the effect would not have occurred either (e.g., Pearl 2000; Sloman
2005; this view can be traced back to Hume 1748/1999). Let us exemplify this with the
following sentence (this is a deterministic sentence, a probabilistic example is provided
further below in this section):
(41) The land was flooded because of the heavy rain.
Elementary s-causation—(7) in Section 3—-can capture this deterministic sentence in
the following way (the content that appears in association with the s-causation elements
has been added in underlined subscript right after the element):
(42) C: Agoland-xdry (–), Antheavy rain-xdiff flooded (+) → E: Agoland-xdiff flooded
This reads: the cause (C) involves a cognized entity land (Ago), cognized to have the
initial action/state value of being dry (x), which is weaker (–) than another cognized
entity heavy rain (Ant), cognized as having an action/state value that is different to the
one of Ago (xdiff, here, flooding). Since Ant (heavy rain) is stronger (+), the effect (E)
of the interaction is that Ago has to take on the value of Ant, hence the land is flooded.
41
If one now were to remove the stronger Ant (the conceptualization of heavy rain) and its
stronger force (the power of flooding), then the resultant would not be different to the
initial state of Ago: the land would be conceptualized as remaining dry, in accordance
with the initial state that is conceptualized in relation to the land (in relation to Ago).
In counterfactual words: In the possible world where the heavy rain had not occurred,
the land would have remained dry. In force-dynamic terms, the possible world where
Ant and xdiff(+) are removed can be represented as something like:
(43) s-causation with removal of stronger Ant
Agoland-xdry → Agoland-xdry
In other words, removing the imposing stronger Ant (heavy rain) with its different
value (flooding) leaves nothing but Ago, the land, with its state of being dry. Since no
Ant intervention takes place, the resultant is that the land remains unflooded. In the
possible world where Ant did not intervene, Ago remains unchanged. This then is how
digital force dynamics can specify the actual mental elements and structures needed to
conduct counterfactual reasoning about causation (cf. Wolff, 2007, p. 104, who develops
a comparable argument for analog force dynamics).
Similarly, force-dynamic elementary causation can also be integrated into proba-
bilistic causation (cf. again Wolff 2007, p. 104). Consider the following example taken
from Wolff (cf. with (41)):
(44) Heavy rains cause flooding.
What one needs to additionally consider in relation to the elements of s-causation—as
formulated in (7)—is a distinction that a causal relationship of an individual occurrence
(such as (41)) is deterministic and a causal relationship of multiple occurrences (such
as (44)) can be probabilistic. A simple solution to this is to treat the individual oc-
currence as the unmarked (default) case and multiple occurrences as the marked case.
42
Thus for s-causation in an individual occurrence—such as (41)—the description in (7)
suffices. In case of multiple occurrences—the marked case—I suggest adding to this
elementary description the concept of probability (prob) as an element that can embed
the entire s-causation structure (the newly added conceptual elements multiplex 6 and
probability have been marked bold in (45b), so that they can be spotted at a glance):
(45) Probabilistic s-causation
a. [C: Ago-x(–), Antmultiplex-xdiff(+) → E: Ago-xdiff]prob
b. [C: Agoland-xdry(–), Antmultiplex heavy rains-xdiff flooded (+) → E:
Agoland-xdiff flooded]prob
(45b) reads: the cause (C) involves a cognized entity land (Ago; which in (44) is not
linguistically mentioned, but must be there for logical reasons) with the cognized initial
value to be dry, which is weaker (–) than another cognized entity heavy rain (Ant)
which is cognized as having the different value of flooding the land. Since Ant (heavy
rain) is stronger (+), the effect (E) of the interaction would be that the land were
flooded. However, if Ant is additionally marked as multiplexed (Antmultiplex) within the
cause (within C), then this results in the entire cause (C) and effect (E) to be embed-
ded in probability. This contrast of uniplex= deterministic / multiplex= probabilistic
reflects the common-sense notion that a generalization over a group of entities might
not necessarily hold true for each individual entity within the group.
This solution to integrating force-dynamic cause and effect into probabilistic causa-
tion is quite different to the one proposed by Wolff (2007) for his account of force dy-
namics. In relation to his model he writes: “In the dynamics model, uncertainty is built6Multiplexity is a term from Talmy (2000b, pp. 48–50). It stands for “conceptual plurality” (rather
than grammatical plurality). Thus chair and furniture are grammatically speaking both singular,while conceptually the former is uniplex and the latter multiplex. Since the basic issue here is “con-ceptual plurality,” the conceptual uniplex-multiplex contrast is considered here rather than the formal(morphologically and syntactically relevant) singular/plural contrast.
43
into the representation of causation because people do not know the exact magnitude of
the vectors” (p. 104). However, this proposed solution is somewhat contradictory, since
Wolff also—rightly, I suppose—assumes that people also do not know the exact mag-
nitude of the force-dynamic vectors for a single occurrence either (pp. 88+98). Given
this, there is nothing in the logic of this reasoning that could prevent the prediction
that a single causal occurrence—for example (41), The land was flooded because of the
heavy rain—could also be probabilistic, which of course it is not. Thus Wolff’s link-
ing of imprecise (subjective) vector magnitudes with the notion of probability leads to
a problem that one could call “overprobability”—the problem that probability can be
wrongly predicted for a deterministic scenario. The solution in (45) does not suffer from
this drawback since probability can only be triggered off by multiplexity and thus is
effectively excluded from the realm of singular (uniplex) occurrence. In sum then, the
current force-dynamic account of cause and effect is, like Wolff’s force-dynamic account,
compatible with counterfactual, deterministic, and probabilistic causation, but unlike
the latter does not face the problem of “overprobability.”
7.5 Force-dynamic elementary causation and the mental time
line
Based on decades of research, Talmy has sketched a comprehensive theory of concept
structuring. Thus far this theory has four main schematic systems: configurational
structure, perspective, attention, and force dynamics (Talmy 2000d). Talmy also pro-
vides a first sketch how force-dynamic elements relate to the other three schematic
systems:
. . . force dynamics is a fourth schematic system: to the preceding basically
pictorial complex [consisting of the schematic systems configurational struc-
44
ture, perspective, attention], one now adds the forces that the elements of the
structural framework exert on each other. While the first three schematic
systems [configurational structure, perspective, attention] relate most di-
rectly to our system of visual perception, force dynamics relates most to
the kinesthetic system (2000a, p. 467).
Stocker (2012a,b) has provided some suggestions how the first three schematic systems
of Talmyan concept structuring—configurational structure, perspective, and attention—
are mentally represented. In this context he has proposed the mental existence of points,
lines, embodied/disembodied perspective points, and the like in an overall framework
of what he refers to as Talmyan concept structuring. One suggestion in Stocker (2012a,
2013) is for instance that the mental time line is not merely a didactic aid that al-
lows us to think about time, but that it is an actual spatial structure in our mind
that allows for mental time travel—for example by projecting one’s mental gaze along
this mentally construed past/future line or by mentally construing one’s whole body
to move along that line (cf. Hartmann & Mast 2012; Merritt et al. 2010; Miles et al.
2010a,b,c, 2011; Ulrich & Maienborn 2010; Ulrich et al. 2012). As a first approximation
toward an integration of mental force-dynamic interactions with the work on the mental
time line in an overall framework of Talmyan concept structuring, one can speculate
that, on the one hand, the elements of a (successful) mental cause—that is, Ago-x(–),
Ant-xdiff(+)—are draped onto one temporal region of this time line and that, on the
other hand, the elements of a (successful) mental effect—that is, Ago-xdiff—are draped
onto a subsequent region on this mental time line. This draping of these two clusters of
force-dynamic elements onto the two temporal regions turns these two temporal regions
into a cause and into an effect.
45
References
Ali, N., Chater, N., & Oaksford, M. (2011). The mental representation of causal con-
ditional reasoning: Mental models or causal models. Cognition, 119, 403–418.
Barbey, A. K. & Wolff, P. (2007). Learning causal structure from reasoning. In Proceed-
ings of the 29th Annual Conference of the Cognitive Science Society (pp. 713–718).
Nashville, TN: Erlbaum.
Deane, P. (1996). On Jackendoff’s conceptual semantics. Cognitive Linguistics, 7(1),
35–91.
Goldvarg, E. & Johnson-Laird, P. (2001). Naive causality: A mental model theory of
causal meaning and reasoning. Cognitive Science, 25, 565–610.
Hartmann, M. & Mast, F. W. (2012). Moving along the mental time line influences the
processing of future related words. Consciousness and Cognition, 21, 1558–1562.
Hume, D. (1748/1999). An enquiry concerning human understanding. New York: Ox-
ford University Press.
Ibarretxe Antuñano, I. (2006). Leonard Talmy. A windowing onto conceptual structure
and language: Part 2: Language and cognition: Past and future. Annual Review of
Cognitive Linguistics, 4, 253–268.
Jackendoff, R. (1990). Semantic structures. Cambridge, MA: MIT Press.
Jackendoff, R. (1996). Conceptual semantics and cognitive linguistics. Cognitive Lin-
guistics, 7(1), 93–129.
Kant, I. (1781/1998). The critique of pure reason. New York: Cambridge University
Press.
47
Kant, I. (1787/2009). Kritik der reinen Vernunft. Köln: Anaconda, 2nd edition.
Merritt, D. J., Casasanto, D., & Brannon, E. M. (2010). Do monkeys think in
metaphors? Representations of space and time in monkeys and humans. Cognition,
117, 191–202.
Miles, L., Betka, E., Pendry, L., & Macrae, C. (2010a). Mapping temporal constructs:
Actions reveal that time is a place. The Quarterly Journal of Experimental Psychol-
ogy, 63(11), 2113–2119.
Miles, L., Karpinska, K., Lumsden, J., & Macrae, C. (2010b). The meandering mind:
Vection and mental time travel. PLoS One, 5(5), e10825.
Miles, L., Nind, L., & Macrae, C. (2010c). Moving through time. Psychological Science,
21(2), 222–223.
Miles, L., Tan, L., Noble, G., Lumsden, J., & Macrae, C. (2011). Can a mind have
two time lines? exploring space–time mapping in mandarin and english speakers.
Psychonomic Bulletin & Review, 18(3), 598–604.
Pearl, J. (2000). Causality: Models, reasoning, and inference. Cambridge, England:
Cambridge University Press.
Sloman, S. (2005). Causal models: How people think about the world and its alternatives.
New York: Oxford University Press.
Sloman, S., Barbey, A., & Hotaling, J. (2009). A causal model theory of the meaning
of cause, enable, and prevent. Cognitive Science, 33(1), 21–50.
Spirtes, P., Glymour, C., & Scheines, R. (2000). Causation, prediction, and search.
Cambridge, MA: MIT Press, 2nd edition.
48
Stocker, K. (2012a). The time machine in our mind. Cognitive Science, 36(3), 385–420.
Stocker, K. (2012b). Toward an embodiment-disembodiment taxonomy. Cognitive
Processing – International Quarterly of Cognitive Science, 13(Suppl 1), S347–S350.
Stocker, K. (2013). Digital causal cognition. International Journal of Cognitive Lin-
guistics, 4(1), 9–34.
Stocker, K. (In Press-c). The theory of cognitive spacetime. Metaphor and Symbol,
[to appear in Vol. 29(2) (2014), estimated advanced online publication date: 31 Mar
2014; estimated print publication date 30 Apr 2014].
Talmy, L. (1976). Semantic causative types. In M. Shibatani (Ed.), Syntax and se-
mantics (Vol. 6): The grammar of causative constructions. New York: Academic
Press.
Talmy, L. (1985). Force dynamics in language and thought. In Papers from the twenty-
first regional meeting of the Chicago Linguistic Society (pp. 293–337). Chicago Lin-
guistic Society.
Talmy, L. (1988). Force dynamics in language and cognition. Cognitive Science, 12(1),
49–100.
Talmy, L. (2000a). Force dynamics in language and cognition. In Toward a cognitive
semantics. Vol. 1: Concept structuring systems chapter 7. Cambridge, MA: MIT
Press.
Talmy, L. (2000b). The relation of grammar to cognition. In Toward a cognitive
semantics. Vol. 1: Concept structuring systems chapter 1. Cambridge, MA: MIT
Press.
49
Talmy, L. (2000c). The semantics of causation. In Toward a cognitive semantics. Vol.
1: Concept structuring systems chapter 8. Cambridge, MA: MIT Press.
Talmy, L. (2000d). Toward a cognitive semantics. Vol. 1: Concept structuring systems.
Cambridge, MA: MIT Press.
Talmy, L. (2011). Cognitive semantics: An overview. In C. Maierborn, K. von
Heusinger, & P. Portner (Eds.), Semantics: An international handbook of natural
language meaning, volume 1. De Gruyter Mouton.
Ulrich, R., Eikmeier, V., de la Vega, I., Ruiz Fernández, S., Alex-Ruf, S., & Maienborn,
C. (2012). With the past behind and the future ahead: Back-to-front representation
of past and future sentences. Memory & Cognition, 40, 483–495.
Ulrich, R. & Maienborn, C. (2010). Left–right coding of past and future in language:
The mental timeline during sentence processing. Cognition, 117(2), 126–138.
Walsh, C. R. & Sloman, S. (2011). The meaning of cause and prevent: The role of
causal mechanism. Mind & Language, 26(1), 21–52.
Wolff, P. (2003). Direct causation in the linguistic coding and individuation of causal
events. Cognition, 88(1), 1–48.
Wolff, P. (2007). Representing causation. Journal of Experimental Psychology: General,
136(1), 82–111.
Wolff, P. & Song, G. (2003). Models of causation and the semantics of causal verbs.
Cognitive Psychology, 47(3), 276–332.
Wolff, P. & Zettergren, M. (2002). A vector model of causal meaning. In Proceedings of
the 24th annual conference of the Cognitive Science Society (pp. 944–949). Hillsdale,
NJ: Erlbaum.
50