Professional Documents
Culture Documents
.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.
BRILL is collaborating with JSTOR to digitize, preserve and extend access to Phronesis.
http://www.jstor.org
Phronesis XLIIIII
GLENN LESSES
It quickly becomes evident that the Stoic view about the natureand
extent of the propositionalcontentof mentalstates is puzzling.They deny
that humanyoung or animalscan entertainbeliefs or have thoughts.The
existence of propositionalattitudesordinarilyis held to involve the attributionof some thoughtor belief. So, it would follow that the Stoics deny
small childrenor animalshave mentalstates with propositionalcontents.
Yet, thereis also evidenceto the contrary.For instance,in his well-known
example, Chrysippusattributeswhat appearsto be reasoningby disjunctive syllogism to a dog. One might argue that this kind of example commits the Stoics to the ascriptionof propositionalcontentseven in the case
of animals. Furthermore,if the mental states of non-rationalanimalsare
so impoverishedas to lack any propositionalcontent, the Stoics must
explain how it is possible for such animalsto functionadequatelyat all.
Doesn't, say, a mouse see that a cat is nearby?Recently,commentators
have debatedthe extent of propositionalcontentsin the Stoic accountof
psychological states. The discussion has focused, in particular,on perceptional states because for the Stoics perceptionand impulse are what
distinguishanimal life from other living things and perceptionhas a special prominencein their inquiryinto mentalstates.Two basic, competing
interpretations
have emerged.On what has become the orthodoxreading,
the Stoics sharply separatethe psychological states of non-rationalanimals from those of rationalanimals.Accordingto advocatesof the orthodox position, the mental states of, say, small childrenare too simple to
have the cognitive structurenecessaryfor the attributionof propositional
attitudesto them.4Non-rationalanimalscan perceiveonly qualities such
scholars of contemporarydiscussions. Here are just two examples. While Julia Annas,
Hellenistic Philosophy of Mind, does not discover the ancestor of any particularcontemporary position in Stoic thought, she does suggest, e.g., pp. 1-2, that the Stoics
construct the first philosophy of mind that is recognizably contemporary. Richard
Sorabji, "PerceptualContent in the Stoics," Phronesis 35 (1990), pp. 307-314, goes
much further. He argues that the Stoics would endorse the basic views of Daniel
Dennett on propositionalattitudesratherthan those of Donald Davidson.
4 Michael Frede, "Stoics and Skeptics on Clear and Distinct Impressions," The
SkepticalTradition(Berkeley:Universityof CaliforniaPress, 1983), ed. Myles Burnyeat,
pp. 65-93, is among the clearest and most forceful proponentsof this view. Others
include Brad Inwood, Ethics and Human Action in Early Stoicism (Oxford: Oxford
University Press, 1985), pp. 73-75, A.A. Long and D.N. Sedley, The Hellenistic
Philosophers, vol. I (Cambridge: Cambridge University Press, 1987), pp. 239-241,
ChristopherGill, "Is there a concept of person in Greek philosophy?,"Companionsto
ancient thought 2: Psychology (Cambridge:Cambridge University Press, 1991), ed.
Stephen Everson, pp. 166-193, and Jean-LouisLabarriere,"De la 'naturephantastique'
the Stoics
as sweetnessor whiteness.Accordingthe second interpretation,
deny thatthe sensoryperceptionsof animalsare completelydevoid of propositional contents. On this alternativereading, the Stoics much more
liberally ascribe propositionalcontent even in the case of non-rational
animals.5
In this essay, I argue that the Stoics draw some distinctions- as they
often are proneto do - that, as a result,enable us to locate a solutionto
the puzzle. The Stoics deny that there are any raw psychological states
completelydevoid of cognitivecontentbecause they discriminatebetween
a conceptionof more robustand more narrowcontent.It follows that the
Stoics occupy an intermediateposition,less severe thanwhat the orthodox
interpretationattributesto them and less generousthan what the alternative account finds. Thus, the Stoics hold that there is considerablecontinuity as well as significantdifferencesbetweenthe psychologicalstates of
rationaland non-rationalanimals.
I
GLENN LESSES
CONTENT,CAUSE,AND STOICIMPRESSIONS
[Sextus,
M 7. 227-231]
Both Cleanthesand Chrysippussharethe view thatan impressionis something physical, namely, a modificationof some pneuma. They disagree
about how to explain this physical change, but both rule out treating
dering might misleadingly suggest that it primarily involves visual images and the
sense-modality of sight. Michael Frede, "Stoics and Skeptics on Clear and Distinct
Impressions,"and A.A. Long and D.N. Sedley, The Hellenistic Philosophers, vol. 1,
employ "impression."Various other attempts to render the term include: Julia Annas,
Hellenistic Philosophy of Mind, and Richard Sorabji, "Perceptual Content in the
Stoics," use "appearance,"Brad Inwood and L.P. Gerson, Hellenistic Philosophy:
IntroductoryReadings (Indianapolis: Hackett Publishing Co., 1988), "presentation,"
(which also has the virtue of neutralitywith respect to sense-modalities, but doesn't
reflect the definition of phantasia in terms of tup6sis quite as well) and A.A. Long,
"Representationand the self in Stoicism," Companions to Ancient Thought 2: Psychology (Cambridge: Cambridge University Press, 1991), ed. Stephen Everson, pp.
102-120, uses "representation."A.A. Long, p. 107, n. 6, changes his usage from his
earlier "impression"to "representation"because he worries that the Humean associations of "impression"might mislead.
12 See D.L. 7. 50, Sextus M 7. 227-231, 7. 372-373, Plut. Ad Col. 1122C, Comm.
not. 1084F.
1' See also D.L. 7. 45-46, 50. Unless otherwise indicated, translationsare based on
Brad Inwood and L.P. Gerson, Hellenistic Philosophy: IntroductoryReadings, with
occasional modifications.
GLENN LESSES
The second stage involved in perception proper is assent (sunkatathesis). For perceptionto occur, an agent must not only be presentedwith
an impression,he or she also mustassentto it.'6In rationalanimals,assent
to an impressionis a voluntaryact.'7The Stoics often speak of sunkata14 For discussions of this passage in Sextus, see Julia Annas, Hellenistic Philosophy
of Mind, pp. 72-75, Deborah Modrak, "Stoics, Epicureans, and Mental Content,"
p. 99, A.A. Long and D.N. Sedley, The Hellenistic Philosophers, vol. 1, p. 239, and
David Sedley, "Chrysippuson psychophysical causality," pp. 329-330. According to
Modrak, the passage suggests that Chrysippus does not accept that an impression is
an imprint. But the text indicates that the dispute between Chrysippusand Cleanthes
over what Zeno meant is instead a matterof interpretingthe natureof a tuposis. Both
appear to accept that an impression is an imprintof some kind.
's Julia Annas, Hellenistic Philosophy of Mind, pp. 74-75, points out Sextus might
not do justice to the dispute. It is possible that Chrysippussimply attemptedto develop
furtherwhat he held to be the same fundamentalview as Cleanthes' position.
However, Chrysippusappears to criticize Cleanthes for accepting a naive view of
mental representationthat implies that perceptualimpressions are copies of the qualities that the objects representedhave. For instance, on Cleanthes' account of impression, when we perceive a triangularobject the impression that we ordinarilyhave is
triangular.Some commentatorshave taken the remarks of Chrysippus to entail that
impressionsare propositionalin form or articulablein linguistic form. E.g., Annas, pp.
74-75, concludes that Chrysippus "analyzed perception in terms of the reception of
content and its articulationin linguistic form." This conclusion is too strong. All that
seems to follow is that the impressionconveys information.Whetherwhat is conveyed
requires propositional content is less clear. Why is it necessary that our perceptual
impressions of, say, squareness or redness, involve the propositionthat something is
red or square? As it stands, all Chrysippus has to accept is that the qualities that
impressions have are not necessarily similar to the characteristicsof their causes.
16 See D.L. 7. 49, Cicero Acad. 1. 40, 2. 145, Plut. Ad Col. 1122B-C, Stobaeus
1. 349. 23-27.
'7 The nature of assent for non-rationalanimals is a vexed and complicated question. Brad Inwood, Ethics and HumanAction in Early Stoicism, p. 72, denies that nonrational animals can assent. Jean-Louis Labarriere,"De la 'nature phantastique'des
animaux chez les Stoiciens," especially, pp. 243-249, argues that the Stoic position
is that animals can exhibit a type of assent to impressions. See also Julia Annas,
Hellenistic Philosophy of Mind, pp. 72-75, 89-102, Christopher Gill, "Is there a
concept of person in Greek philosophy?,"pp. 185-186, A.A. Long and D.N. Sedley,
The Hellenistic Philosophers, vol. 1, p. 322, A.A. Long, "Representationand the self
thesis resultingin some sort of belief. Much dependson the natureof the
impressionto which agents are assentingsince assent is a sort of genus.
To outlinethe basic Stoic position,what agentsassent to, at least in many
cases, is a proposition(lekton)characterizedby the impression.Genuinely
conceptualthinkingarises subsequentlyto the occurrenceof impressions.'8
The Stoics hold thatconceptsare a type of impression,which resultsfrom
an agent, in effect, internalizingrepeated,similar impressions.'9The act
of thinking involves articulatingor reflecting about the proposition to
which one has assented.Althoughthe Stoics distinguishtheoreticallybetween the occasion of having an impressionand assentingto it, it is unclearwhetherthetwo stagesareactuallydistinctin cases of sense-perception.
II
GLENNLESSES
What is Frede's main evidence?Text (2) leads Frede to hold that any
impressionof a rationalbeing is rational.
(2) Of impressions, some are rational (logikai) and some are non-rational (alogoi).
The rational are those of rational animals, the non-rationalof non-rational.The
rational,then, are thoughtsand the non-rationalhave been given no special name.
[D.L. 7. 51]
Through the senses alone one is not able to grasp the truth, as we indicated
before23and now shall explain briefly, for they [the senses] are by nature nonrational, and of more than being impressed by impressors [i.e., the things that
impress] they are not capable, as they are completely unsuitablefor discovering
the truth.For not only must one be moved to have a sensation of white or sweet
for one to grasp the truth in the underlying things, but one must be brought to
have an impression of that thing that "this is white" and "this is sweet." And so
the other like things [i.e., the other senses]. But to grasp a thing of this kind is
no longer the work of perception.For color only and taste and sound is its nature
to grasp, while that "this is white" or "this is sweet," which are neithercolor nor
taste, is unsuspected by sense.24[Sextus M 7. 344-345]
CONTENT,CAUSE,AND STOICIMPRESSIONS
For Frede, (3) indicatesthat the senses by themselvesare unable to produce impressions containing any propositionalcontent. This task, accordingto (4), requirescertainoperationsof the mind. Thus, rationaland
non-rationalanimals differ in the sorts of impressions of which they
are capable. Frede explains the capacity of rationalanimals to entertain
rationalimpressionsin terms of their ability to form concepts. Lacking
any conceptualapparatus,non-rationalanimals cannot entertainimpressions having propositionalform. In the Stoic account of psychological
development,concepts (ennoiai) arise later than perceptualimpressions
(Aetius 4. 11. 1-5). The Stoics separateconceptualthinkingfrom the mere
occurrenceof perceptualstates.
Although Frede's formulationof the Stoic account has considerable
plausibility,RichardSorabjirejects the orthodoxreading. He challenges
both the philosophicalunderpinningsof Frede's interpretationas well as
its textual support.Let us first consider Sorabji's objection to a philosophical argumentemployed by Frede and next describe his own fresh
interpretiveproposal.Sorabjiis unimpressedby the theoreticalargument
that conceptsare necessaryfor propositionalthought.It is a controversial
matterfor manycontemporaryphilosophersof mindwhetherin every case
the employment of concepts is requiredfor propositionalattribution.26
2' This translation basically follows A.A. Long and D.N. Sedley, The Hellenistic
Philosophers.
26 See Richard Sorabji, "PerceptualContent in the Stoics," pp. 308-309, "Animal
10
GLENN LESSES
11
12
GLENN LESSES
13
D.L. 7. 86.
14
GLENN LESSES
15
16
GLENN LESSES
Whatis the natureof the initialimpressionsof self-perception?Answering this questionwill help determinethe extent of the Stoic attributionof
propositionalcontents.Since these impressionsoccurfrom the momentof
birth,the initial impressionscannot be the result of repeatedexperience.
Stoic examplesof self-perceptionoften referto specific animalbody parts
and functions.For instance,snails perceivetheirflesh and shells and birds
perceive theirwings are for flying.5'Self-perceptionsometimesis said to
involve perceivingthat somethingis the case but since such descriptions
are far from typical we should exercise cautionabout drawingtoo much
from this evidence.52The Stoics often seem to be searchingfor a way of
speakingthat is less cognitively loaded and appearto be reluctantto attributetoo much cognitionto animalsand humaninfants.53In any event,
these kinds of examples - such as the bull perceivingthat its horns are
for self-defense - appeal to later stages of oikeiosis. But self-perception
also occurs duringthe initial stages of oikei6sis.
What about the originalimpressionsthat arise in self-perceptionin the
CONTENT,CAUSE,AND STOICIMPRESSIONS
17
Neitherof these passagesprimarilyconcernsthe kindsof examplesof selfperceptionprovidedby Hierocles such as the bull's awarenessof its own
horns for self-defense.The accountshere are generaldescriptionsof any
impressionof sense. In addition,text (6) speaks of the impressionsas the
source of any cognition, which strongly suggests that these descriptions
pertainto the earliest stages of oikeiosis.
First, a preliminaryquestion must be addressed. Do these passages
apply to both non-rationaland rationalanimals?55There is no reason to
I
The language of (6) is so similar to (5) that we should suppose that Antiochus is
speaking in his Stoicizing mode and that the passage genuinely reflects Stoic doctrine.
S5 This question is raised by Jean-LouisLabarrire, "De la 'naturephantastique'des
18
GLENN LESSES
actuallyperceivedby the mind,not the senses - the propositionthatsomething is white. There are puzzles about this passage to which we will
returnshortly,but for the momentat least it is reasonableto suggest that
the sense-impressionhas the impressedthing, viz., the white, as partof its
contentsbut does not have the propositionthatsomethingis white as part.
Passages (5) and (6) also stipulatethat a sensoryimpressionis capableof
conveying some additionalcontentfor we are told that the sense impression reveals itself. Thus, provisionallywe can say that sense impression
of white includes the following contents:(i) the white, which indicates
the cause of the impression,and (ii) the occurrentimpression(i.e., of the
white) itself. At this point, we do not have to commitourselvesto specifying that these mentalcontentsare propositionalin form. But it is clear
that the propositionthat somethingis white goes beyond the ascriptionof
these modest contents.
Accordingto (5), the impressionof white is then the basis of our saying that"thereexists somethingwhite."The Stoics are cautioushere about
animaux chez les Stoiciens," pp. 238-243, though in the context of whether animals
have any self-consciousness. As a result, he never directly poses the question about
the cases of self-perceptionon which we focus.
19
the source of that propositionalinformation.Is the propositionthat something white exists alreadycontainedin the impression?If it is, then it is
temptingto conclude that non-rationalminds are incapableof discerning
all the contents already containedin the impression.But an alternative
reading is also possible. It must be conceded that the evidence underdetermineseither interpretation,however, the second reading is preferable
for the sake of consistencywith our earlieranalysis of (5) and (6). These
sources are fairly specific aboutwhat narrowcontentsare containedgenerally in impressions.In addition,(4) tells us that rationalanimals bring
other mental operationsinto play which affect their impressions.On the
preferredreading, a rational mind contributes some new information,
which expresses the propositionthat somethingis white, to the contents
A propositionalmentalstate of this sort
of the originalsense-impression.56
might well be broadlydescribedas inferential.A rationalanimal is capable of inferringthat, e.g., if a white thing causes my impression,then it
is the case that somethingis white.For the Stoics, any inferenceaboutthe
cause of the impressionof whiteness requiresan ability of which nonrationalminds are incapable.Even in the well-knownChrysippeanexample, the dog employing disjunctive syllogism is only said to simulate
reasoning.57One might plausibly hold that inferentialreasoning, however minimaland automaticthe inference,requiresotheroperationsof the
mind besides perception.
Let us returnto text (4), which makes some distinctionaboutcontent.58
The senses are said to perceivethingssuch as whiteness.Yet, Ciceroholds
that the senses perceive only in a way propositionalclaims such as that
somethingis white. It is not altogetherobvious how to unpackthis passage.59If we read Cicero ratherliterally here, he distinguishesbetween
56
Julia Annas, Hellenistic Philosophy of Mind, pp. 82-84, discusses a parallel set
of interpretationsfor kataleptic vs. non-katalepticimpressions in rational animals.
S7 Sextus, PH 1. 69 states that the dog in effect (dunamei) is reasoning. Richard
Sorabji, "PerceptualContent in the Stoics," p. 313, takes this instance of simulated
reasoning to be more or less a legitimate case of reasoning.
58 I am making a weaker point here than Frede, who argues that text (4) rules out
the possibility of non-rationalimpressions having content at all.
59 We saw that the passage, according to Sorabji, "PerceptualContent in the Stoics,"
p. 311, commits the Stoics to impressions having propositional form. He proposes
that if the senses can perceive in a way that something is the case then in a way
such impressions can have propositional form. He states that to be "propositionalin
a way" when the senses perceive that, say, this is white can be understood as the
senses "presenta non-verbalizedappearanceas of something's being white."
20
GLENN LESSES
what the senses can perceive only in a way and what the mind - here,
referring to mental operations in the hegemonikon other than senseperception- perceives fully. Perceptioninvolves assent so that in each
instance the mind assents to what is presented.On the orthodoxview,
there is simply less to assent to in the case of what is presentedonly to
the senses. There also appearsto be a readingconsistentwith Sorabji's
interpretationif what the mind contributesis an articulationof the content that is already present in the original impressionof sense. Let me
suggest a thirdpossibility.If animalminds are incapableof makinginferences from what they do in fact perceive,then Cicero's remarksthat the
mindalone securesthat, say, this is white appliesonly to rationalanimals.
The senses of rationalanimals alone can grasp in a way that something
is white because they alone can draw inferencesbased on their sensations. Frede's orthodox interpretationis mistaken in supposingthat (4)
supportsthe view that animalsperceivesonly qualities.Sorabji'salternative is also incorrectbecause non-rationalimpressionsare not the issue
here. The context of Acad. 2. 21 shows that the impressionsbeing consideredare only those of rationalanimals.So Cicero's discussionapplies
only to impressionsof rationalanimals and has little to do with impressions of sense generally.Y0
In texts (5) and (6), we saw that the impressionsof sense are said to
reveal their causes. In the case of an impressionof whiteness, the cause
of the impressionis the white, which is the impressedthing (phantaston).
The passagesspeakof "thewhite"(to leukon)or "thevisible"(to horaton),
which could refer, of course, either to the quality of an object or to an
object itself. But we do not need to commit ourselves to either alternative and so will speakjust of qualitiesin what follows. Additionally,the
claim that the impression'scause, the white, is revealed might be taken
in two ways. First, one might hold that the impression reveals something, which, as it happens,caused it. In the case of the impressionof a
white thing,what gets revealedis just whiteness.If all the Stoics mean is
that the perceiversees whiteness, then nothingmore is containedin the
impressionthan the rudimentaryperceptionof a quality,whiteness.This
reading of the texts is consistent with the orthodoxinterpretation.But,
what gets revealedwhen an
accordingto a second possible interpretation,
impressionrevealsits cause is considerablymorecontent-laden.The white
is revealedas the cause of the impression.So, when these passages say
61
21
61
My renderingclosely follows Brad Inwood and L.P. Gerson, Hellenistic Philosophy, here. A.A. Long and D.N. Sedley, The Hellenistic Philosophers, vol. 1, translates the passage "[the impression]must be capable of revealing the self-evident object
that caused it," which even more strongly suggests that more than just a quality is
revealed. A self-evident feature is that the quality or object is related to the impression as its cause.
62 Nothing in this discussion is meant to suggest that such impressionsare kataleptic.
22
GLENN LESSES
Very generally,oikei6sis describesa processof somethingcoming to belong. If applied to individuals,the process refers to that which comes to
"belong to me" or is "of mine."'-6 The account of mental contents we have
describedthus fits well with oikeiosis. The Stoics hold that in the early
stages of oikei6sis impressionsoccur because all animals from birth on
come equippedwith impulsesand the basis of theirinitialimpulsesis selfperception.These impressionsinclude as part of their contents that the
impressionitself belongs to the perceiver.In this minimalsense of selfperception,the Stoics attributesufficient cognitive functioningto nonrationalanimals so as to discriminatebetween their own psychological
states and whatever is alien64to them. The contents of the impressions
themselves indicate that they are the perceiver'sand not of something
independent.65
An example might help to illustratethe basic featuresof mentalcontent that have emerged in our discussion. Suppose that Dion has a dog,
Zeno. Whatare the contentsof Zeno's sensoryimpressions?If Zeno sees
his paw, he has an impressionwhich, first of all, he is aware of as his.
In addition,variouscausal relationsconstitutepartof the contentsof his
impression.Zeno's impressioncontains at least somethinglike the following: his impressionis causedby his paw. Thus, it can be said of Zeno
that he perceives his paw. What cannot be attributedto Zeno is that he
perceives that it is a paw. The Stoics also deny that Zeno draws inferences from what he perceivesor that he can assess the truthof the contents of his impressions.The basic point is that the Stoics are making
some sort of distinction between perceiving and perceiving that. They
want to explain why Zeno perceives his paw but doesn't perceive that
it is a paw. They construct their explanation in terms of a somewhat
restrictedset of propositionalcontents.
23
These problemscould have led the Stoics to offer a more familiar theory of causal
contents in which the meaning of the impressionis simply whatever causes the impression. This kind of content does not require attributingany propositionalcontent at all
to non-rationalanimals. Such a position would then be consistent with the orthodox
view. But there is little evidence that they do, in fact, propose a version of this sort
of causal theory of content.
67 On the Stoic view, it is clear that rational animals have impressions that are
robustly propositional in form. Roughly, beliefs involves assent to an impression
which contains what are termed"sayables"(lekta). Sayables are rathersimilar to statements. Earlier,we saw that impressions containing sayables are equivalent to thoughts
and occur only in rational beings.
24
GLENN LESSES
" Sorabji evidently tries to formulate something weaker than full-fledged propositional content when he employs his conception of predicationalcontent. Though he
never completely explains his notion, it enables him to distinguish between predicational content and the strongertype of content involved in assenting to a belief. However, he at least must clarify why saying of an animal's impression that, for instance,
the scent is in a certain direction commits him to so much less in terms of propositional contents than the position that there is an impression that something is white.
There obviously are complex issues of reference and opacity to be dealt with in both
cases, which we will put aside.
See A.A. Long and D.N. Sedley, The Hellenistic Philosophers, vol. 1, p. 240, for
one attempt to explain a position that at first appears similar to Sorabji's. They discuss this sort of impression in terms of an example. Imagine seeing a John Wayne
movie in which "we get the impression of John Wayne's being on the screen in front
of us, but not of course the impression that John Wayne is on the screen in front of
us (p. 240)." Their point is really to distinguish a person having beliefs from merely
having impressions. For Long and Sedley, only the second is a case of belief. However, they do not explicitly deny the first case also has propositionalcontent. Their
example shows we must exercise caution about specifying the propositionscontained
in an impression, not that there is none. One might argue that the first case also
involves beliefs, though there are issues about what is believed. Long and Sedley deny
that non-rational animals can have concepts and agree with the orthodox view that
non-rational impressions are empty of mental content. But this illustrationconcerns
questions about a person's impression of Wayne's being on the screen. Hence, issues
about whether their account commits us to denying the ascriptionof impressionswith
propositionalform to non-rationalanimals never really arise.
69 The project undertakenin this essay is limited to constructing an interpretation
that is intermediatebetween the two standardones. There are a numberof open questions. Perhapsthe most importantis how the two-stage model of perceptionapplies to
impressions having only modest mental contents. The natureof assent in the case of
narrow content must be distinguished from that of robustly propositionalimpressions.
How exactly does the mind assent in the former case? There are related questions
about the nature of impulses since perception is the basis of impulse. Any impulses
to which non-rationalimpressions give rise must be differentin kind from the impulses
generated by rational impressions. Ultimately, these issues can affect Stoic moral
theory because the Stoics tie their account of moral development to the nature of
impulse.
25
70 My initial work on these issues began while I was a visiting scholar at Harvard
University, where I was generously allowed the use of its resources. I have delivered
differentparts of early versions of this essay to a numberof audiences to whom I am
grateful, especially, to those present at the American Philosophical Association meeting in Chicago in April 1995. I have benefited from suggestions and criticisms on all
these occasions, and, in particular,from the subsequent written comments of Shaun
Nichols and Joyce Carpenter.I owe a special debt to John Ellis, the respondentat the
APA meeting, whose stimulatingdiscussion and written remarksgreatly improved this
essay.