332 R. M. SHAPLEY AND C.

ENROTH-CUGELL

Therefore, in the original FH model, the relation differential equations for the BHL model, up to the
between gain and time to peak of the impulse z~ stage is:
response is a power law where the exponent is one
less than the number of stages which are affected
dy,(t)
by adaptation.
- I(t) - a y,(t) (31)
dt
6.1.2. THE BAYLOR -- HODGKIN -- LAMB MODEL
dyj(t)
In the treatment of turtle cone responses, Baylor
- - - a [Yj_l(t) - y~(t)]
et al. (1974b) modified the original F H model by dt
setting

a, = as = a = 1 / T . (29) dy.(t) dz,(t)

- a Yn ~(t) -
dt dt
Several other kinetic schemes were found to fit the
waveforms of cone responses better (Baylor et al., K21
1974a) but the modified FH model was chosen {K,2 + v z,(t)}z,(t) + {1 + v Z2(t)} z2(t)
adequate for an expanded theory of adaptation and A
responses too large to be in the linear range. It also
has a much simpler interpretation than the F H dz2(t)
model as a chain of chemical reactions (compare - g,~ {1 + v z2(t)} z,(t) -
Borsellino et al., 1965, with Baylor et al., 1974b). dt
Figure 59 illustrates the model used by Baylor et al. {K,2 [1 + vZ2(t)] + K23}Z~(t).
(1974b), which we will call the B H L model.
There are important differences between the BHL In these equations v is the autocatalytic constant
model and the original F H model. Since, in the implicitly defined by the equation K,2 = /(,2 + vz2.
B H L model, the rate constants for buildup and A is the equilibrium constant of zx, z2 and is equal
decay of the y substance are the same, there is no to K , 2 / K 2 , . This is a complicated set of differential
change in the gain when the value of the rate equations and an intuitive feeling of what is going
constants is changed, as there is in the FH model. on is difficult. Some insight may be gained by
This can be seen by considering the impulse examining a special case of interest: a weak,
response of the first n stages of the B H L model: sinusoidally modulated increment which produces
a response in the linear range. In this case,
y , (t) = Q • B • ( t / T ) " - ' e t/T (30)

where T = 1 / a , and B is a constant.

In this case the peak of Yn is at (n-1)T and the
magnitude of the peak is independent of T. Gain
control cannot be achieved in this model simply by
an increase of the rate, a. Therefore, the B H L
model has to have a new feature, an extra pathway
for decay of the final stage of the cascade where
the decay rate depends on the level of illumination.
This is shown in Fig. 59 as the pathway denoted z,,
z,, z3, z,. z, is equivalent to Yn and is assumed to
be the concentration of "blocking particles" which
lead to closing of the sodium channels in the
photoreceptor's membrane. The decay of z, to the
next state, z2, is subject to autocatalysis by the level
of z2 present (see Fig. 59). Thus, the system of
/(t) = Io + I, e i~t
yj(t) = Yjo + Yil ei°~'
z~(t) = Z~o + z~le ~t. (32)
We make the approximation that •0>>•, and that
Io, the mean level of illumination which may also
be viewed as the mean arrival rate of light quanta,
is large compared to the rate constants of the
cascade. Then we obtain [cf. equation (40) of
Baylor et al. (1974b)] for the cone's frequency
response in the light adapted range: