Page 165 - Contributed Paper Session (CPS) - Volume 4

CPS2165 Carla Susete Gonçalves Francisco et al.
classified as description of implementation at the same process level,
which the RT theory explains at a higher level. The advantages and
disadvantages of the DDM model are:
• The DDM model provides a direct mechanism to account for and
predict the probabilities of delays and their latencies.
• Suggestive approach to the behavior of neuronal populations.
Presence of auditory fluctuations. Complexity of the DDM model.
Doubts about the linearity of the accumulation process:
experimental observations suggest following some kind of
exponential law.
LATER (Linear Threshold Approximation with Ergodic Rate) Model. LATER, is a
model originally derived empirically and vulnerable to experimental testing.
Over the last decade we have been attempting to verify this functional
interpretation by trying to challenge its three elements:
• S0 represents log prior probability. The change in reaction time is
linearly related to the log probability.
• ST represents a threshold. The main part swivels about a fixed
intercept. Reduction in latency is associated with a large increase in
the number of early responses.
• µr represents the supply of information. The rate of information
supply affects the mean rate of rise of the decision signal, and this
turn causes the distributions not to swivel, but to be shifted
horizontally in a parallel fashion. We consider that r is a Gaussian
2
variable with mean µ and variance .
• Time between the start and the threshold is:
T=(ST – S0)/r
• Reciprocal of latency, 1/T:
1/T=r/(ST – S0)
• Following that r is a Gaussian random variable, distribution of 1/T is
Gaussian with mean /( −0) and variance /( − 0) .
2
2
All biological systems are subject to unpredictable perturbations,
technically known as noise. the sensorial noise does not contribute
significantly to the variability of reaction time, according with a large
number of evidences that strongly suggest that, in most conditions, -
randomness (Signal-to-noise ratio in sensory systems). Neurophysiological
experiments show that the contribution of randomness to the overall
variability of reaction time is insignificant. Suggestions such:
• LATER decision mechanism can be thought of as being preceded by
a detection stage, obeying random-walk dynamics.
• The observed randomness of reaction time does not originate in the
outside world, it is deliberately injected into the system from within.
Agents try to be as unpredictable (random) as they possibly can.

154 | I S I W S C 2 0 1 9