Communication based on private states

From: Harrison, Richard (
Date: Mon Oct 23 1995 - 11:33:05 BST

Some comments on the Lubinski & Thomas paper.

l&t> Figure 1 illustrates our animal model: A pair of birds was trained to
l&t> exchange arbitrary cues, "letters," based on drug-induced state
l&t> variations in the internal environment of one of them. The drug-cue
l&t> bird received cocaine, pentobarbitol, or saline and was trained to
l&t> communicate with another bird by pecking response keys corresponding to
l&t> these drug-induced state variations.

Is this communication? If in Stage B (in the actual figure -
internal state reporting) a random key was pecked then the Decoder
would still match correctly and be rewarded (presumably due to prior
reinforcement of this behaviour). Wouldn't a definition of
communication involve an intent by the initiator to influence the
response of the receiver. This situation does not seem to involve any
such intent.

l&t> The tacter's deprivation and satiation conditions were alternated
l&t> (ABAB) and the birds' communicative performance was observed.
l&t> When satiated and without visual access to the mander, the tacter
l&t> stopped matching letters to colors and this condition was eventually
l&t> terminated. Subsequently, however, the opaque barrier was removed for a
l&t> final condition (with the tacter deprived and satiated in alternate
l&t> ABAB sessions, as before). With visual access to the mander reinstated,
l&t> the tacter's matching behavior reappeared, following the mander's
l&t> request, even though the tacter was food and water satiated. (Lubinski
l&t> and MacCorquodale, 1984)

Seems to be the first indication in the article of emergent
communication in pigeons that doesn't leave open the possibility that
reinforced behaviours are simply being combined independent of

l&t> Although not responsible for all subjective components of emotional
l&t> states, the circulatory, digestive, proprioceptive, and respiratory
l&t> systems are involved in providing interoceptive stimulation on which
l&t> statements about affects are based; and covariations in the properties
l&t> of these physiological events as a function of reinforcing stimuli,
l&t> emotionally charged punishers and reinforcers, are well documented
l&t> (Buck, 1987; Carlson, 1986; Tuma & Masur, 1985).

Does this mean a change in these interoceptive stimuli changes

l&t> After completing a component of the interlocking sequence, each bird
l&t> gradually began to orient toward the stimulus change in the adjacent
l&t> chamber, which was in close proximity to the other bird. After
l&t> consuming food or water, for example, the drug-cue bird approached the
l&t> area near the decoder's "How do you feel?" key. If the decoders were at
l&t> all sluggish in pecking the key when this light became illuminated, the
l&t> drug-cue bird would rapidly peck the Plexiglas directly above the key
l&t> while orienting toward the decoder.

This behaviour certainly seems to have some intention behind it (NB.
When I wrote my first comment I didn't know L & T had made this
observation). Nevertheless, the decoder does not understand anything
from the interaction about the drug-cue bird from the 'interaction'. The most
the aggresion implies is;
      "Press you're key, I'm thirsty."
It seems unlikely that the Drug-cue bird has communicated (in any
successful sense) knowledge about it's internal state to the Decoder.

l&t> Even though such clients may have good reason to feel angry, they are
l&t> unable to report that feeling. In this situation, the therapist, or
l&t> possibly other members of a therapy group, gesticulate, raise their
l&t> voices, or in other ways present an aversive setting to
l&t> client-listeners until they blurt out their feelings.

This a weak analogy to the pigeon model. How many human interactions
involve people who have no interest in expressing themselves. Even if
you accept that the above scenario as one such example, it differs
from the pigeon model in that the therapist or group understands (at
least to some extent) the internal sate the client is reporting.

l&t> The human tendency to conceal emotional stimulation for social
l&t> advantage makes it understandable that such language based repertoires
l&t> would be more common in humans than in other organisms.

Does the causal link between 'human tendancy to conceal emotional
stimulation' and the commonality of a language-based repetoire humans
display necesserily work in the direction L & T suggest?


This archive was generated by hypermail 2b30 : Tue Feb 13 2001 - 16:23:55 GMT