April 1999
CO230: COGNITIVE SCIENCE

QUESTION 1 (Compulsory)

Total Marks: 20 Marks

 Click here to access other questions

GRADE A
Sample student's solutions are indicated in green.
Return to Question 1

 
(a) In the information processing approach, knowledge can be defined as declarative or procedural. Briefly describe these terms and give an example of each. [4]
Declarative knowledge:
  • Knowledge that represents fact, i.e knowledge about "What is....."
  • For example : 911 is known to seek for help when emergency occurs.

Procedural knowledge:

  • knowledge about the procedures to be operated on those fact representation i.e knowledge about "How to....."
  • For example : To dial 911, firstly I need to find a telephone; secondly dial 9 followed by 1 and another 1. This is about how to operate on the fact 911.

 
(b) Using the following predicates:
  • birds(X)
  • animals(X)
  • has_feathers(X)
  • can_fly(X)

express the following statements as quantified predicate expressions:
(i) All birds are animals. [1]
"X : Dx birds(x) ® animals (X)

 
(ii) If an animal has feathers then it is a bird. [1]
"X : Dx [ animals(X) ^ has_feathers(X)]  ® birds (X)

 
(iii) Some birds can fly and all birds have feathers. [2]
[ $X : Dx birds(X)  ® can_fly(X) ] ^ [ "X : Dx birds(X)  ® has_feathers(X) ]

 
(c) In fuzzy logic, single inference makes use of a single inference rule to find output for given input values. Give an example of a fuzzy rule (briefly explaining why it is fuzzy). Describe the three steps used in a single inference. [4]
Example: If a person is tall, he is heavy. Here, two linguistic variables are used: tall, heavy. They are inexact and vague.

The three steps :

  1. Fuzzification : work on the input set to get a membership value.
  2. Inference : work on the output set to get a membership value.
  3. Defuzzification : based on the resulting set, calculate the result.

 
(d) Briefly describe how an artificial neuron works in terms of the inputs, weights activation function, and output. [4]
  1. All inputs signals are multiplied by its associated weights.
  2. The weights are the connections between neurons. The weights may strengthen the input signal or weaken the input signal.
  3. All weighted signals are set to the summation block which adds up all weighted values and send the resulting value NET to an activation function.
  4. The activation function maps input into output.
  5. Then a threshold function is applied to produce an output signal. For example, if a step threshold function is used, the output is 1, if the NET is greater than the threshold; the output is 0, if the NET is less than or equal to the threshold.

 
(e) (i) What does the imitation game designed by Alan Turing test? [2]
- Alan Turing's imitation game tests whether or not a computer has intelligence.
- In the game, an invigilator tests a human being and a computer with attempt to finding their identity.

 
(ii) What are the flaws in the rationale of this test? [2]
-It is a fundamental mistake trying to restrain computer intelligence in human mould.