action-patterns), of acquired habits (handwriting, spoken accent),
and in the stereotyped routines of thought; the integrative tendency
is reflected in flexible adaptations, improvisations, and creative
acts which initiate new forms of behaviour.
4.7 Under conditions of stress, the self-assertive tendency is manifested
in the aggressive-defensive, adrenergic type of emotions, the
integrative tendency in the self-transcending (participatory,
identificatory) type of emotions.
4.8 In social behaviour, the canon of a social holon represents not only
constraints imposed on its actions, but also embodies maxims of
conduct, moral imperatives and systems of value.
5. Triggers and Scanners
5.1 Output hierarchies generally operate on the trigger-releaser principle,
where a relatively simple, implicit or coded signal releases complex,
preset mechanisms.
5.2 In phylogeny, a favourable gene-mutation may, through homeorhesis
(Waddington) affect the development of a whole organ in a harmonious
way.
5.3 In ontogeny, chemical triggers (enzymes, inducers, hormones) release
the genetic potentials of differentiating tissues.
5.4 In instinctive behaviour, sign-releasers of a simple kind trigger off
innate releasive mechanisms (Lorenz).
5.5 In the performance of learnt skills, including verbal skills,
a generalized implicit command is spelled out in explicit terms on successive
lower echelons which, once triggered into action, activate their
sub-units in the appropriate strategic order, guided by feedbacks.
5.6 A holon on the n level of an output-hierarchy is represented on the
(n+1) level as a unit, and triggered into action as a unit. A holon,
in other words, is a system of relata which is represented on the
next higher level as a relatum.
5.7 In social hierarchies (military, administrative), the same principles
apply.
5.8 Input hierarchies operate on the reverse principle; instead of triggers,
they are equipped with 'filter'-type devices (scanners, 'resonators',
classifiers) which strip the input of noise, abstract and digest its
relevant contents, according to that particular hierarchy's criteria
of relevance. 'Filters' operate on every echelon through which the
flow of information must pass on its ascent from periphery to centre,
in social hierarchies and in the nervous system.
5.9 Triggers convert coded signals into complex output patterns.
Filters convert complex input patterns into coded signals. The former
may be compared to digital-to-analogue converters, the latter to
analogue-to-digital converters. [22]
5.10 In perceptual hierarchies, filtering devices range from habituation
and the efferent control of receptors, through the constancy
phenomena, to pattern-recognition in space or time, and to the
decoding of linguistic and other forms of meaning.
5.11 Output hierarchies spell, concretize, particularize. Input hierarchies
digest, abstract, generalize.
6. Arborization and Reticulation
6.1 Hierarchies can be regarded as 'vertically' arborizing structures
multiplicity of levels and form 'horizontal' networks: arborization
and reticulation are complementary principles in the architecture
of organisms and societies.
6.2 Conscious experience is enriched by the cooperation of several
perceptual hierarchies in different sense-modalities, and within
the same sense-modality.
6.3 Abstractive memories are stored in skeletonized form, stripped of
irrelevant detail, according to the criteria of relevance of each
perceptual hierarchy.
6.4 Vivid details of quasi-eidetic clarity are stored owing to their
emotive relevance.
6.5 The impoverishment of experience in memory is counteracted to some
extent by the cooperation in recall of different perceptual hierarchies
with different criteria of relevance.
6.6 In sensory-motor coordination, local reflexes are short-cuts on the
lowest level, like loops connecting traffic streams moving in opposite
directions on a highway.
6.7 Skilled sensory-motor routines operate on higher levels through networks
of proprioceptive and exteroceptive feedback loops within loops,
which function as servo-mechanisms and keep the rider on his bicycle
in a state of self-regulating, kinetic homeostasis.
6.8 While in S-R theory the contingencies of environment determine
behaviour, in the present theory they merely guide, correct and
stabilize pre-existing patterns of behaviour (Weiss).
6.9 While sensory feedbacks guide motor activities, perception in its turn
is dependent on these activities, such as the various scanning
motions of the eye, or the humming of a tune in aid of its auditory
recall. The perceptual and motor hierarchies are so intimately
cooperating on every level that to draw a categorical distinction
between 'stimuli' and 'responses' becomes meaningless; they have
become 'aspects of feed-back loops' (Miller et at).
6.10 Organisms and societies operate in a hierarchy of environments,
from the local environment of each holon to the 'total field',
which may include imaginary environments derived from extrapolation
in space and tune.
7. Regulation Channels
7.1 The higher echelons in a hierarchy are not normally in direct
communication with lowly ones, and vice versa; signals are
transmitted though 'regulation channels', one step at a time.
7.2 The pseudo-explanations of verbal behaviour and other human skills
as the manipulation of words, or the chaining of operants, leaves
a void between the apex of the hierarchy and its terminal branches,
between thinking and spelling.
7.3 The short-circuiting of intermediary levels by directing conscious
attention at processes which otherwise function automatically, tends
to cause disturbances ranging from awkwardness to psychosomatic
disorders.
8. Mechanization and Freedom
8.1 Holons on successively higher levels of the hierarchy show increasingly
complex, more flexible and less predictable patterns of activity,
stereotyped and predictable patterns.
8.2 All skills, whether innate or acquired, tend with increasing practice
to become automatized routines. This process can be described as the
continual transformation of 'mental' into 'mechanical' activities.
8.3 Other things being equal, a monotonous environment facilitates
mechanization.
8.4 Conversely, new or unexpected contingencies require decisions to be
referred to higher levels of the hierarchy, an upward shift of
controls from 'mechanical' to 'mindful' activities.
8.5 Each upward shift is reflected by a more vivid and precise
consciousness of the ongoing activity; and, since the variety of
alternative choices increases with the increasing complexity on
higher levels, each upward shift is accompanied by the subjective
experience o
f freedom of decision.
8.6 The hierarchic approach replaces dualistic theories by a serialistic
hypothesis in which 'mental' and 'mechanical' appear as complementary
attributes of a unitary process, the dominance of one or the other
depending on changes in the level of control.
8.7 Consciousness appears as an emergent quality in phylogeny and
ontogeny, which, from primitive beginnings, evolves towards more complex
and precise states. It is the highest manifestation of the
integrative tendency to extract order out of disorder, and
information out of noise.
8.8 The self can never be completely represented in its own awareness,
nor can its actions be completely predicted by any conceivable
information -- processing device. Both attempts lead to infinite
regress.
9. Equilibrium and Disorder
9.1 An organism or society is said to be in dynamic equilibrium if the
self-assertive and integrative tendencies of its holons counter-balance
each other.
9.2 The term 'equilibrium' in a hierarchic system does not refer to
relations between parts on the same level, but to the relation
between part and whole (the whole being represented by the agency
which controls the part from the next higher level).
9.3 Organisms live by transactions with their environment. Under normal
conditions, the stresses set up in the holons involved in the
transaction are of a transitory nature, and equilibrium will be
restored on its completion.
9.4 If the challenge to the organism exceeds a critical limit, the balance
may be upset, the over-excited holon may tend to get out of control,
and to assert itself to the detriment of the whole, or monopolize its
functions -- whether the holon be an organ, a cognitive structure
(idée fixe), an individual, or a social group. The same may happen
if the coordinating powers of the whole are so weakened that it is
no longer able to control its parts (Child).
9.5 The opposite type of disorder occurs when the power of the whole over
its parts erodes their autonomy and individuality. This may lead
to a regression of the integrative tendencies from mature forms of
social integration to primitive forms of identification and to the
quasi-hypnotic phenomena of group psychology.
9.6 The process of identification may arouse vicarious emotions of the
aggressive type.
9.7 The rules of conduct of a social holon are not reducible to the rules
of conduct of its members.
9.8 The egotism of the social holon feeds on the altruism of its members.
10. Regeneration
10.1 Critical challenges to an organism or society can produce degenerative
or regenerative effects.
10.2 The regenerative potential of organisms and societies manifests itself
in fluctuations from the highest level of integration down to
earlier, more primitive levels, and up again to a new, modified
pattern. Processes of this type seem to play a major part in
biological and mental evolution, and are symbolized in the universal
death-and-rebirth motif in mythology.
APPENDIX II
AN EXPERIMENT IN PERCEPTION*
Arthur Koestler and James J. Jenkins
The writers are indebted to Donald Foss for collecting and coding the data.
Thanks are also due to Professor Douglas Lawrence and Professor Ernest
Hilgard of Stanford University and to Professor Arnold Mechanic and
Joanne D'Andrea of California State College at Hayward for their generous
facilitation of the study.
ABSTRACT
Experience suggests that a common error in processing visual sequences is inversion or transposition of two or more adjacent items. This phenomenon suggests that information concerning the identity of items and their positions may be partially separable. A perception experiment was performed with tachistoscopic exposure of 5-, 6-, and 7-digit sequences. Abundant evidence was found for transposition errors. Further, such errors were distributed in a serial position curve much like that found for errors of single items.
* See Chapter 1, 13, and p. 297. Reprinted with permission from
Psychon. Sci., 1965, vol. 3, pp. 75-6.
PROBLEM
While information-processing in visual perception has received increasing attention in recent years [1], one common phenomenon of faulty processing which may have some theoretical significance seems to have been ignored. We refer to the inversion (or transposition) of adjacent items in a sequence of numbers shown in a tachistoscope. Though such errors are common enough in bookkeeping and have earned a special proofreader's mark, they are absent from discussions of visual perception or memory span in standard works such as Osgood [2] and Woodworth and Schlosberg. [3]
Apprehending a series of numerals and subsequently repeating them in their correct sequence must either involve the ordered storage of the individual items or the storage of information relating to that order. Both information identifying an item and information defining its place in the sequence must be available for the S <subject> for successful performance of the task.
The potential separability of the two kinds of information involved is not easy to demonstrate. If a subject makes a single error of identity, reporting either an incorrect number or a blank, it may indicate that he has lost only identity information. This argument, however, is inconclusive, because if the subject had acquired no information at all regarding the offending item, but complete information regarding other items, the outcome would be the same. The inversion of two digits or the permutation of three or more digits, on the other hand, furnishes a compelling argument because it is prima facie evidence that the identity information is accurate while the positional information is incomplete or distorted.
The purposes of the present study were to demonstrate that the phenomenon of transposition could be observed under laboratory conditions and to describe the locus of its probable occurrences in a given sequence.
METHOD
The stimulus materials were 80 4x6 notecards upon which digit sequences were typed in elite type. The 80 sequences were divided into four sets of 20 cards each. The first set showed sequences 5 digits in length; the second and third sets showed 6-digit sequences; the fourth set showed 7-digit sequences. The sequences contained the digits 1-9 with never more than a single digit repeated on a given card. The repeated digit, if any, never occurred without at least one intervening digit. The sets were presented in the order given above. A random arrangement was made of each set. This arrangement was used in the forward order for half the Ss and in reverse order for the remainder. The materials were presented in a mirror-type tachistoscope.
The Ss were 14 undergraduates in introductory psychology courses. The S held a plunger switch which activated the tachistoscope. The E gave a ready signal when the stimulus card was in place. The S activated the tachistoscope when he was ready. He was instructed to say the digit sequence aloud immediately after its appearance, and was encouraged to guess if he was not sure of one or several items. The S always knew how many digits were shown. Responses were recorded on a tape recorder. Only one exposure per sequence was given and the S was not given any information about the correctness of his response.
Two practice sequences with ascending limits were given to accustom the S to the apparatus and to provide the E with some information on threshold. The test sets were then presented. One-minute rest periods were given after each set.
Exposure duration was individually adjusted for each S. Pilot work suggested that transpositions occurred most readily at the point wher
e the S was beginning to miss single digits in the sequence. Therefore, the E attempted to have the exposure interval long enough that the proper number of digits would be reported but short enough that they were not always reported with complete accuracy. After every five cards the E decided whether to keep the exposure the same or to change it. Since there were practice effects in the task and since the task became appreciably more difficult, E continued to modify the presentation time during the course of the experiment. Generally 10-msec steps were employed in such changes but with an occasional S whose performance was markedly inferior the step span was increased.
RESULTS AND DISCUSSION
Responses were transcribed from the tape and scored. The following categories were employed:
C -- correct
E -- gross error
I -- one digit incorrect, or 'blank' reported for a single missing digit
T -- transposition of adjacent pairs of digits with rest of
sequence correct
T1 -- transposition of three or more digits with remainder correct
IT -- transposition of two or more digits and one digit incorrect
O -- other errors, usually experimental or equipment errors
Results are given in terms of these scoring categories in Table I. Examination of the table shows that transposition provides an important source of errors. It is, however, difficult to find a statistical model which would provide a precise evaluation of the statistical significance of such errors. As Woodworth and Schlosberg [4] point out in their discussion of scoring memory span, any scoring system which attempts to provide separate credit for accuracy and order is arbitrary. Thus, any statistical model must make assumptions about the S's strategies on the one hand (e.g., Did the S note that digits can repeat within a sequence and, if so, did this alter his guessing behaviour in the appropriate manner?) and the interrelationships of error types (which we do not yet know) on the other. Fortunately, the question is not crucial for present purposes. The only question that need be asked here is whether there is more transposition than would be expected by chance (however chance is to be defined).