SKINNER (1935)- Two types of conditioned reflex and a pseudo type

TWO TYPES OF CONDITIONED REFLEX AND A PSEUDO TYPE[*] 

From the Biological Laboratories of Harvard University

B.F. Skinner (1935)[1]

First published in Journal of General Psychology, 12, 66-77.

A conditioned reflex is said to be conditioned in the sense of being dependent for its
existence or state upon the occurrence of a certain kind of event, having to do with the
presentation of a reinforcing stimulus. A definition that includes much more than this
simple notion will probably not be applicable to all cases. At almost any significant
level of analysis a distinction must be made between at least two major types of
conditioned reflex. These may be represented, with examples, in the following way
(where S = stimulus, R = response, (S - R) = reflex, -> = "is followed by," and [  ] = "the
strength of" the inclosed reflex):

Given such a sequence,

where [S1-R1] ~ = 0,[2]
conditioning occurs as a
change in [So-Ro] - an
increase in strength
(positive conditioning) in
(a) and a decrease (negative
conditioning) in (b).[p.67]

Given the simultaneous or

successive presentation of
S'o and S'1, where [S'1- R'1] ~ =0,[2] conditioning occurs as an increase in [S'o - R'1].
Differences between the types are as follows:

1. In Type I, So ->Ro-> S1, where Ro necessarily intervenes between the stimuli; in Type
II, S'o ->S'1, where R'o is ignored.[3] In I, Ro is important; it becomes the conditioned
response. In II, R'o is irrelevant (except when it is relevant in another sense by
conflicting with R'1) and may actually disappear.

Since conditioning of the second type may take place even when S'1 occurs after R'o,
Paradigm II, Example (c), may be written for this case as follows:

light --------------- r ---------------> food ---------------salivation,

when it is identical with I. But the result is not to reduce the two types to a single form.
Both kinds of conditioning proceed simultaneously but separately. If r is "turning
toward the light," for example, and if the food is withheld until turning takes place,
[light - turning] will increase according to I while [light-salivation] will increase
according to II. The same result is obtained with negative conditioning. Example (d)
gives, upon delaying S'1,
light --------------- r ---------------> shock --------------- flexion, etc.,

where [light--r] will decrease according to I, while [light-flexion] increases according to

II.
In the special case in which Ro and R'1 are of the same form, the two kinds can
apparently not be separated. If, for example, some unconditioned salivation is supposed
to be elicitable by a light,[4] we may substitute "salivation" for r, to obtain

light - salivation I ---------------> food - salivation II.

Both [light - salivation I] and [light - salivation II] will increase, with apparently no
possible distinction. This is a very special case and is also in no sense a reduction to a
single type.

2. In I, (So - Ro) occurs normally in the absence of (S1 - R1), [p.68] and its strength may
be measured without interfering with the reinforcing action of S1. In II S'1 must be
withheld whenever a measurement of [S'o - R'1], the conditioned reflex, is taken, because
S'1 also evokes R'1. Some amount of extinction necessarily ensues in the second case.

3. Since S'1 must be withheld in measuring [S'o - R'1], R'1 must be independent of any
property of S'1 not possessed by S'o. In Example (c) salivation may become attached to
the light as a conditioned response of Type II; but seizing, chewing, and swallowing,
which are also responses to S'1, must not be included in the paradigm since they require
the presence of parts of S'1 which cannot be supplied by S'o.

A special restriction on Paradigm II is therefore necessary. Where S'o is of a very simple

sort (a tone, for example), the properties possessed in addition to S'o by S'1 are practically
equal to S'1, and we may express the restriction in terms of a general distinction between
two kinds of response. The first kind require no external point of reference in their
elicitation or description. Typical examples are: glandular activities (salivation), local
muscular responses (flexion, wink, breathing movements, production of sounds), and
facilitation and inhibition.[5]

The second kind require points of reference for their elicitation or description which are
not supplied by the organism itself, but by the stimulus. Examples are: orientation
toward the source of a sound, approaching a light, and touching, seizing and
manipulating objects (such as a lever or food). Our present rule is that responses of the
second kind cannot be substituted for R'1 in Paradigm II, unless S' also supplies the
required points of reference.

4. In Type I, [So - Ro] ~ = 0 before conditioning takes place. The reflex-to-be-

conditioned must be elicited at least once as an unconditioned "investigatory" reflex. In
Type II, [S'o - R'1] may begin at zero and usually does. In Type I the state of the reflex is
"conditioned" by the occurrence of the reinforcing sequence, but its existence is not. A
distinction between a conditioned and an unconditioned relfex is here less significant,
because all examples of the [p.69] former have necessarily been examples of the latter.
There are no exclusively conditioned reflexes in this type.

Since [S'o - R'1] may begin at zero, a new reflex may be created in conditioning of the
second type. And since practically any stimulus may be attached to R'1 in Paradigm II, a
very large number of new reflexes can thus be derived. Conditioning of Type I, on the
other hand, is not a device for increasing the repertory of reflexes; Ro continues to be
elicited by the one stimulus with which it began.

There are three reflexes in Paradigm II, but only two in I.

5. The significant change in Type I may be either an increase or a decrease in strength;

in Type II it is an increase only, even when [S'o - R'1] does not begin at zero.

In Type I stimuli may be divided into two classes, positively and negatively
conditioning, according to whether they produce an increase or decrease when used as
reinforcement. The distinction cannot be made in Type II, where a reflex may be
negative in another sense (a reflex of "avoidance," for example), but where its strength
only increases during conditioning.

6. In Type I the conditioned reflex (So - Ro) may be associated with any drive; in Type II
the reflex (S'o - R'1) is necessarily attached to the drive specified by R'1.

This point may require some comment. In the present use of the term a drive is an
inferred variable of which the strength of a group of reflexes is a function (2). Hunger,
for example, is a variable (H) a change in which is responsible for concurrent changes in
the strength (a) of all unconditioned reflexes concerned with the ingestion of food, (b)
of all conditioned reflexes (of either type) in which the reinforcing stimulus is
concerned with the ingestion of food, and (c) to a much lesser extent of all
"investigatory" reflexes. In Paradigm I, Example a (lever - pressing) is originally a
function of H to some slight extent under (c) above. After conditioning it varies with H
according to (b), over a wide range probably equal to that of any unconditioned reflex
under (a). Conditioning of Type I is really the becoming attached to a group of reflexes
varying as a function of some drive. This is a much more comprehensive description of
the process than to define it as an increase in strength, where the drive is assumed to
remain constant at a significant value. But the identity of H in the present case is [p.70]
determined only by our choice of a reinforcing reflex. Given (S1 - R1) of another drive,
say thirst, then (So - Ro) will become conditioned by attaching itself to the group varying
with thirst, and will not vary with H except to some slight extent under (c).

This is a characteristic wholly lacking in Type II. Here R' is originally part of the
unconditioned reflex and the drive to which it belongs is definitely fixed.

7. A minor difference is in the way in which the stimulus-to-be-conditioned usually

acts. In Type I, So is usually part of a larger field, and Ro occurs as the result of the
eventual prepotency of So over other stimuli. In Type II, S'o is usually suddenly presented
to the organism. The significance of this difference, which is not absolute, will appear
later.

We shall now consider a third type of relation which involves a discrimination. It may
be based upon a conditioned reflex of either type, but we shall begin with I. To establish
a discrimination subdivide So into two classes on the basis of a selected property or
component member (3, 4). For example, let the lever stimulate either in the presence of 
a light (L), when the stimulus may be written as SA B..L..(subcripts indicate properties or
components), or in the dark, when the stimulus is SA.B.... Continue to reinforce the re-
response to one of them, say SA.B..L.., and extinguish or negatively condition the response
to the other by breaking the sequence at S1 or by introducing as S1 of the negatively
conditioning kind (Difference 5). When this has been done, [SA.B..L.---Ro] > [SA.B..---Ro].
And at any value of the underlying drive such that (SA.B..L ---Ro) is usually elicited but
(SAB ---Ro) is not, there exists the following condition. Given an organism in the presence
of SAB.. ordinarily unresponsive, the presentation of L will be followed by a response.
For the sake of comparison we may set up a paradigm in imitation of II as follows:

The relation between the light and the response

to the lever might be called pseudo-
conditioned reflex. It has some of the
characteristics of Type II: the original response to the light is irrelevant [p.71]
(Difference 1); the relation may be wholly absent prior to the "conditioning" (Difference
4); it changes in a positive direction only (Difference 5); and the "stimulus" is usually of
the presented kind (Difference 7). In all these respects it differs from Type I, although
the example is based upon a reflex of that type. I many other respects it differs from
both types. A reinforcing reflex is not included in the paradigm, but must be added as a
third or fourth reflex. The response is not principally to the light, but to the lever; the
light is only a component member of the whole stimulus, and "light-pressing" is not
legitimately the expression of a reflex. The lever cannot be removed to show the
conditioned effectivelness of the light as in Type II; instead, the response to the lever
alone must be extinguished - a characteristic that we have not met before.

In spite of these differences it is often said (in similar cases) that the light becomes the
"conditioned stimulus for the response to the lever" just as it becomes the stimulus for
salivation. This is a confusion with Type II which obviously arises from a neglect of the
extinguished reflex. The relation of pressing the lever to the lever itself is ignored and
only the relation to the light taken into account. The lever comes to be treated, not as a
source of stimulation, but as part of the apparatus, relevant to the response only for
mechanical reasons. When the discrimination is based upon a response not requiring an
external point of reference (Difference 3), the chance of this neglect increases
enormously. If we substitute "flexion of a leg" for "pressing a lever" (and continue for
the moment with Type I), So in Paradigm I is not directly observable; we simply wait
until a flexion appears, then reinforce. Having established (So - Ro) as a conditioned
reflex of some strength, we subdivide our inferred So as before, extinguish (SAB..- Ro), and
reinforce (SAB..L..- Ro). When the discrimination has been set up, we have a condition in
which the organism is ordinarily unresponsive but immediately responds with flexion
upon presentation of the light.

Our inability to demonstrate So makes it difficult to show the discriminative nature of

this relation; but it is by no means impossible to find other grounds, as we may see by
comparing it with a true reflex of Type II. Let the presentation of the light be followed
by a shock to the foot until the light alone elicits flexion. The resulting reflex is
superficially similar to the relation of light and [p.72] flexion that we have just
examined, but fundamentally the two cases are unlike. Assuming that no immediate
difference can in fact be detected,[6] we may still show differences by referring forward
or backward to the history of the organism. The two relations have been established in
different ways and their continued existence depends upon reinforcement from different
stimuli. The discriminative drive relation also varies with an arbitrarily chosen drive,
while the conditioned reflex is necessarily attached to the drive to which shock-flexion
belongs.

These differences are chiefly due, however, to the use of a conditioned reflex of Type I
in setting up the discrimination. In a pseudo-conditioned reflex based upon Type II the
distinction is much less sure. Here we are invariably able to neglect the extinguished
member because R'1 is of the kind not requiring an external point of reference
(Difference 3), and we can minimize its importance in other ways. Given a conditioned
reflex of this kind:

if we establish a discrimination between the

tone and the tone-plus-a light (reinforcing the
resonse to the latter), we obtain the following
condition: an organism in the presence of the tone, ordinarily unresponsive, will
respond upon presentation of the light. The only difference between this relation and a
true reflex of Type II is the extinction of the response to the tone, which is evidence that
a discrimination has taken place. The reinforcement of tone and light should condition
responses to both of these stimuli; but we observe that the organism is unresponsive in
the presence of the tone alone.

Now, this surviving difference may be reduced at will by reducing the significance of
S'o in the basic reflex of the pseudo-type. If we lower the intensity of the tone or choose
another stimulus of a less important kind, we may approach as closely as we please to a
conditioned reflex of Type II. We cannot actually reach Type II in [p.73] this way, but
we can easily reach a point at which our pseudo-reflex is identical with any actual
experimental example of that type. This is true because some amount of discrimination
is practically always involved in cases of Type II. When we put a dog into a stand,
present a light and then food, the food reinforces not only the light but the stimulation
from the stand. Merely putting the dog into the stand again should elicit salivation
according to Paradigm II. In practice this is a disturbing effect, which must be
eliminated through extinction. So long as it occurs, any actual case of Type II must be
formulated as a pseudo-conditioned reflex. If SG  is the stimulation affecting the
organism in addition to So, then So Paradigm II should read SG + So. The effect upon SG  is
extinguished through lack of reinforcement in the absence of So, and the result is a
discrimination: an organism is the presence of SG, ordinarily unresponsive, responds
when So is added. The importance of this criticism will depend upon the relative
magnitudes of SG and So. In the optimal experiment SG may be reduced to a value that is
insignificant in comparison with ordinary values of So.

The partially discriminative nature of Type II is invevitable. It is not important in Type I

because of Difference 1. Paradigm I contains an implicit specification that So is active or
has just acted at the moment of reinforcement, since it specifies that S1 is to be withheld
until Ro has occurred. The reinforced stimulus is really So and not SG + So (it is the lever,
in our example, not the whole stimulating field presented by the apparatus). Paradigm II
contains no specification of the activity of So ; and the reinforcing action of S'1 must be
supposed to extend to SG as well as to So. In practice an active state at the moment of
reinforcement is usually insured by presenting So suddenly.[7] This might be included as
an additional provision in Paradigm II, but the provision really required is that So, and
no part of SG, be active at the moment of reinforcement. This is not easily arranged. We
cannot wholly avoid the generalized action of the reinforcement in Type II because of
the lack of dependence of S'1 upon R'o.

One characteristic of the pseudo-conditioned reflex is the variety [p.74] of the forms of
its "stimulus". We have assumed that in our two fundamental paradigms any stimulus
had ultimately the dimensions of energy (although we have often used the shorthand
device of speaking of the source of the energy - as, for example, "lever"). In the pseudo
type, however, the "stimulus" can be a single property. It can be the intensity of the
stimulus, or some such qualitative aspect as pitch or hue. It can be a change from one
value of a property to another, or the absence of a property, or a duration. The reason
why this is possible is that the other properties of the stimulus can be relegated to SG  for
extinction. If the pitch of a tone is to be a conditioned "stimulus," the tone itself must
first become one also, and the response to its other properties must be extinguished by
extinguishing the responses to tones of other pitches. In a true conditioned reflex this
cannot be done. Although it is common to speak of properties as stimuli (1), the
presence of a property in the position of a stimulus is a certain indication that a pseudo-
conditioned reflex is really in quesiton. A property alone cannot be used in either true
type because it implies extinction; most of the real stimulus must be relegated to SG, and
the requirement that the value of SG  be negligible cannot therefore be satisfied.

The position of a pseudo-conditioned reflex may be summarized as follows. When the

pseudo-reflex is based upon a reflex of Type I and when Ro requires external points of
reference, there are important practical and theoretical reasons why a separate
formulation is demanded. When Ro does not require external points of reference, there
are fewer differences, but a separate formulation is still necessary. When the pseudo
type is based upon a reflex of Type II, the distinction is weakened but should still be
made, except when SG can be reduced to a very low value relative to S'o. In the last case
a practical distinction is impossible, not because of an identity of types, but because of
the failure of Type II to appear experimentally in a pure form.

It is a tempting hypothesis that II is not an authentic type but may be reduced to a

discrimination based on Type I. But this has not been shown; we have not reduced the
pseudo type to Type II or vice versa. Nor have we come very near it. The present
pseudo-reflex which resembles II most closely requires of that type for its
establishment. It is probably more than a coincidence that [p.75] a discrimination based
upon Type I has so many of the properties of II, but the reduction to a single type
appears from our present evidence to be highly improbable, desirable though it would
be as an immense simplification. The differences that we have noted are not easily
disposed of. Still more improbable is a reduction of I to II, since the first step supplied
by the pseudo type is then lacking.

To the differences we have listed might be added differences in the parts played by the
two types in the economy of the organism. The essence of Type II is the substitution of
one stimulus for another, or, as Pavlov has put it, signalization (1). It prepares the
organism by obtaining the elicitation of a response before the original stimulus has
begun to act, and it does this by letting any stimulus that has incidently accompanied or
anticipated the original stimulus act in its stead. In Type I there is no substitution of
stimuli and consequently no signalization. Type I acts in another way: the organism
selects from a large repertory of unconditioned reflexes those of which the repetition is
important with respect to certain elementary functions and discards those of which it is
unimportant. The conditioned response of Type I does not prepare for the reinforcing
stimulus, it produces it. The stimulus-to-be-conditioned is never in any sense incidental.

Type I plays the more important rôle. When an organism comes accidentally[8] upon a
new kind of food, which it seizes and eats, both kinds of conditioning presumably
occur. When the visible radiation from the food next stimulates the organism, salivation
is evoked according to Paradigm II. This secretion remains useless until the food is
actually seized and eaten. But seizing and eating will depend upon the same accidental
factors as before unless conditioning of Type I has also occurred - that is, unless the
strength of the reflex (food-seizing) has increased. Thus while a reflex of Type II
prepares the organism, a reflex of Type I obtains the food for which the preparation is
made. And this is in general a fair characterization of the relative importance of the two
types. As Pavlov has said, conditioned stimuli are important in providing saliva before
food is received, but "even greater is their importance when they evoke the motor
component of the complex reflex of nutrition, i.e., [p.7 6] when they act as stimuli to the
reflex of seeking food." (1, p.13).[9] Although "the reflex of seeking food" is an
unfortunate expression, it refers clearly enough to behavior characteristic of Type I.

Footnotes

[*] Accepted for publication by Carl Murchison of the Editorial Board and received in
the Editorial Office, June 4, 1934.

[1] Society of Fellows, Harvard University.

[2] This expression specifies the presence of some amount of drive (2).

[3] For convenience we shall omit the case of simultaneous stimuli in Type II.

[4] See Difference 4 below for this general requirement in Type I.

[5] Where conditioned facilitation and inhibition are defined by substituting for R'1 in
Paradigm II the expressions "Incr.
[S" - R"]" and "Decr. [S" - R"]" respectively. (Incr. = "increase in"; Decr. = "decrease
in").

[6] This is a generous assumption since some evidence for the presence of So can usually
be found. A difference in the character of the response might also be shown (in the case
of the true reflex it may be accompanied by changes in breathing rate, for example,
which would be lacking in the pseudo reflex).

[7] This is our explanation of Difference 7. Another explanation might be added. If S'o is
active for any length of time prior to S1 it will have an extinguishing effect. This cannot
be said of Type I.

[8] That is to say, as the result of weak investigatory reflexes.

[9] This is a doubly interesting statement because Pavlov has confined his own
investigations practically exclusively to conditioned reflexes of the second type. It ought
to be said that he usually regards this type as adequate for the whole field. Thus he says
that the "function of the hemispheres" is signalization (1, p.17), although signalization
is, as we have seen, a characteristic of Type II only. 

References

1. Pavlov, I.P. Conditioned reflexes (Trans. & ed. by G.V. Anrep) London: Oxford
Univ. Press, 1927. Pp xvi + 430.

2. Skinner, B.F. Drive and reflex strength: I. J. Gen. Psychol., 1932, 6, 22-37.

3. ----------. The rate of establishment of a discrimination. J. Gen. Psychol., 1933, 9,

302-350.

4. ----------. The generic nature of the concepts of stimulus and response. J. Gen.
Psychol., 1935, 12, 40-65.

Harvard University
Cambridge, Massachusetts
 

DEUX TYPES DU RÉFLEXE CONDITIONEL ET UN PSEUDO-TYPE

(Résumé)

On définit deux types principaux du réflexe conditionnel et donne une liste des
différences entre eux. Un pseudo-réflexe conditionel, où il s'agit d'une discrimination,
peut être basé sur le Type I pour donner quelques-unes des propriétés du Type II. Quand
it est basé sur le Type II il se peut dans un cas spécial, qu'on ne puisse le distinguer de
ce type, mais cela est dû au manque du Type II de se montrer expérimentalement dans
une forme pure et n'indique pas une réduction à un seul type. Dans une pseudo-réflexe
le "stimulus" peut avoir la forme d'une seule propriété, mais cela n'est possible dans l'un
ni l'autre vrai type. On note la place de chaque type dans l'économie de l'organisme.

Skinner 

ZWEI TYPEN DES BEDINGTEN REFLEXES UND EIN PSEUDOTYPUS

(Referat)

Zwei Haupttypen des bedingten Reflexes werden bestimmt und Unterschiede zwischen
ihnen angegeben. Ein pseudobedingter Reflex, der eine Unterscheidung voraussetzt
kann auf Typus I beruhen, so dass er einige der Eigenschaften von Typus II aufweist.
Wenn er auf Typus II beruht, kann er in einem besonderem Fall nicht unterscheidbar
von jenem Typus sein, aber dies ist die Folge der Nichterscheinung des Typus II
experimentell in einer reinen Form und beweist eine Zurückführung auf einen einzelnen
Typus nicht. Bei einem Pseudoreflex kann der "Reiz" die Form einer einzelnen
Eigenschaft besitzen, aber dies ist nicht möglich in den beiden wahren Typen. Die
Stellung jedes Typus in der Ökonomie des Organismus wird bemerkt.

Skinner