Inferences of innocence

A commonly advanced ground of appeal against conviction is that the evidence supported an inference of innocence and the judge did not correctly tell the jury how they should deal with inferences.

We have passed through a period when much attention was given to what inferences are and how they should be handled. Things got rather complex as efforts were made to be precise about this. The relationship between the process of drawing inferences, and the process of deciding whether something has been proved, was at the centre of this complexity.

Some facts are proved by direct evidence: for example, a witness says that he saw something happen. Other facts are inferred from direct evidence: the facts that are directly proved suggest that something else is true. In turn, these inferred facts may combine, with other inferred facts, or with directly proved facts, or with a mixture of inferred facts and direct facts, to enable a further inference that something else is true. Wherever an inference occurs, it is usually described as an ordinary process of logical thought.

As to proof, it is relatively easy to see that evidence of directly observed facts may prove those facts, to the necessary standard. If the fact is an element of the alleged offence, then, when the jury considers all the evidence in the case, it must decide whether that element has been proved beyond reasonable doubt. But, in relation to other facts, (and with some exceptions), no particular standard of proof is required. The jury does not have to be instructed by the judge about the standard to which they must be satisfied that inferred facts are proved before they can use them to support other inferences, such as an inference of an element of the offence, as long as, ultimately, the jury tests the proof of the elements of the offence against the standard of beyond reasonable doubt.

So, what about the inferences supporting innocence? In what has been regarded as the leading case on inferences in New Zealand, R v Puttick (1985) 1 CRNZ 644 (CA) the Court summarised the position:

"Inference is simply one of the mental processes which may be used by a jury in carrying out its primary task of assessing the evidence and deciding whether or not it establishes the guilt of the accused beyond reasonable doubt. Where the charge has several essential elements, proof of guilt necessarily involves proof of each of those elements to the same standard. It does not, however, require proof beyond reasonable doubt of every fact which may be relevant to proof of each essential element.
… It must be equally unhelpful to tell jurors that, if proven facts support two inferences of equal weight, they should accept one and reject the other. To draw an inference either way from such facts would be pure speculation. Jurors should not be directed to accept or reject inferences when they have no logical basis for either step."

This could suggest (although we can be sure the Court did not intend this reading) that an inference supporting innocence can neither be accepted nor rejected if it is equal in weight to an inference of guilt.

A clearer account has been given in R v Seekamut 10/7/03, CA82/03:

"If on an objective basis that has regard to all the circumstances, a rational alternative to guilt is not excluded, there must for that reason be a reasonable doubt. But the mere fact that some of the circumstances might arguably permit an inference inconsistent with guilt is not enough. The jury’s function is to assess the whole of the evidence and in so doing may conclude that a suggested alternative is not reasonably tenable."

Similarly, yesterday the Privy Council in Taylor v R (Jamaica) [2006] UKPC 12 (13 March 2006), para 18, held:

"Their Lordships agree with the submission made on behalf of the appellant that in the circumstances of this case it was essential that the judge … spell out the possible inferences to be drawn … and instruct them that they must rule out all inferences consistent with innocence before they could be satisfied that the inference of guilt has been proved correct."

This could be read as suggesting a slightly different approach to that in Seekamut: instead of treating the evidence as all being in a big pool, to be considered in the round to see what ends up being proved, the process alluded to in Taylor seems to be one of deciding first whether inferences of innocence can be ruled out, then, if they are, turning to see whether guilt has been proved. We might, however, reasonably wonder whether this explication is correct, insofar as it seems to place a burden of proof on the defence, and it also seems to make an artificial distinction between stages of the reasoning process. The correct approach would be to tell the jury that, if after considering the evidence, they are left with a reasonable doubt about the accused's guilt, they must find him not guilty.

Because the judge in Taylor failed properly to direct the jury in this regard, the appeal against conviction was allowed and the case remitted to determine whether there should be a retrial.