Availability Bias
This bias refers to the fact that when you rely on heuristics, or rules of thumb, to make a decision or form an opinion, there is an inclination to remember more strongly those events that are recalled the easiest. For example, if I were to ask someone whether there were more annual deaths from homicide or stroke in the US, they may very well guess homicide. This could be because of the frequency with which the media reports on these stories and the lack of publicity received by deaths from stroke. In fact, there are eleven times the number of strokes than homicides in the US. The fact that the answer is dictated, even somewhat, by the "availability" of information for mental recall is what is known as the availability bias.
Representativeness
This is essentially stereotyping. An example can be shown through a set of studies performed by De Bondt and Thaler (1985, 1987) showing that stocks that stocks that have lost a lot of money for the past three years tended to outperform stocks that made a lot of money the past three years. However, according to De Bondt (1992), analysts' long-term earnings forecasts tended to have biases in the same direction as recent success or failure. In other words, there was optimism with regard to the analysis of recent outperformers and pessism in the analysis of recent underperformers. There was a lack of consideration of "regression to the mean." Regression to the mean is a statistical concept explaining how values, over time, tend to regress towards their historical average.
Ref: Beyond Greed and Fear: "Understanding Behavioural Finance and the Psychology of Investing" Oxford University Press